DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 11/05/2025 has been entered.
Response to Arguments
Applicant's arguments filed 11/05//2025 have been fully considered by the examiner.
Regarding the rejection of claim 18 under 35 USC 112(b), upon further consideration of the amended limitations, where the problematic language has been removed, the rejection made in the previous office action has been withdrawn.
The remaining remarks are directed to amended language not examined by the examiner, see the current office action below.
Priority
Applicant claims benefit of a prior-filed US Provisional Application No. 63562357 is acknowledged by the examiner.
Drawings
The drawings were received on 02/14/2025 and 4/25/2025. These drawings are acceptable.
Specification
The amendments to the specification filed 04/24/2025 has been entered.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 11/14/2025 is being considered by the examiner.
Response to Arguments
Applicant's arguments filed 11/05/2025 have been fully considered.
Regarding the rejection of claims under 35 USC 112(b), upon further review of the claim limitations, where the problematic language has been removed, the rejection made in the previous office action is withdrawn.
Regarding the rejection of claims under 35 USC 103, the remarks are directed to limitation that have not been examined, see current rejection regarding amended claim limitations.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claim 1, the limitations “generating, based on the measured at least one response, an input for the trained statistical model, wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages; processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model,…” renders the claim indefinite because of ordinary skill in the art is unable to ascertain the intended scope of the claim limitations.
Specifically, the limitations are problematic for the following rationale:
The term “the ANN” has insufficient antecedent basis: it can refer to any artificial neural network or the recited Biological and artificial neural network compressing an BNN which is not the same as a typical artificial neural network without electrode arrays, or something else.
Regarding claim 20, the limitations “measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns;…” renders the claim indefinite because of ordinary skill in the art is unable to ascertain the intended scope of the claim limitations.
Specifically, the limitations are problematic for the following rationale:
The term “the at least one temporal sequence of stimulation patterns” has insufficient antecedent basis, there is no prior recitation of at least one temporal sequence of stimulation patterns.
Regarding claim 19, the limitations are similar to claim 1 and are rejected under the same rationale.
Regarding dependent claims 2-18, the limitations do not resolve the deficiency noted above, in claim 1, and are rejected under the same rationale.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al. (US 20240386258, hereinafter ‘Yang’) in view of Peter Aaser (NPL: “Towards Making a Cyborg: A Closed-Loop Reservoir-Neuro System” hereinafter ‘Peter’).
Regarding independent claim 1, Yang teaches a method for using a biological and artificial neural network (BANN) system to perform a task, the BANN system comprising: (i) a multi-electrode array (MEA); (ii) a biological neural network (BNN) comprising neurons arranged on the MEA, (iii) a trained statistical model; and (iv) at least one processor, the method comprising: using the BANN system to perform: receiving an input signal to be processed by the BANN in furtherance of performing the task; (claimed system including system depicted in Fig 3 and Fig. 4:
PNG
media_image1.png
734
506
media_image1.png
Greyscale
PNG
media_image2.png
434
428
media_image2.png
Greyscale
Fig. 4
And in [0047] FIG. 3 illustrates the use of a neuronal cell culture 100 as a reservoir 300 in reservoir computing. Conventional reservoir computing is a machine learning technique that processes information generated by dynamical systems using observed time-series data. Reservoir computing is a computation framework used for information processing and supervised learning… The reservoir 300 is typically a random or structured dynamical system that is driven by the input signal and has a large number of internal degrees of freedom. The reservoir 300 acts as a “black box” that transforms the input data into a more complex representation that captures the temporal dynamics of the signal. The output data is then fed into a simple readout mechanism that is trained to map the reservoir state to the desired output… [0051] Neuronal cell cultures 100 are well suited to function as a natural reservoir to be combined with conventional computing techniques because they provide a very large number of interconnected nodes—the neural cells. As a neuronal cell culture 100 reacts to a stimulus [(i) a multi-electrode array (MEA); (ii) a biological neural network (BNN) comprising neurons arranged on the MEA] and by utilizing the complex electro-physiological activities already present in these cells, input signals can be mapped into higher or lower dimensional spaces. In order to serve as a compute substrate, such as a reservoir 300 for reservoir computing, neural cell cultures do not necessarily need to be organized into specific structures… [0054] In the implementation illustrated in FIG. 3, the reservoir 300 includes the neuronal cell culture 100 [(i) a multi-electrode array (MEA); (ii) a biological neural network (BNN) comprising neurons arranged on the MEA,] as well as a trainable learning layer 302 implemented in software or a specially constructed circuit. With this configuration, the internal synaptic behavior serves as a high-dimensional substrate for information processing and the output signals from the neurons are provided to an electronic computing device that processes those outputs through the trainable learning layer 302. The trainable learning layer 302 may be implemented as a single layer of an ANN. In this implementation, the output from the trainable learning layer 302 becomes the final output data from the reservoir 300. In some implementations, the trainable learning layer 302 may be implemented biologically with a separate neuronal cell culture [(iii) a trained statistical model]… [0058] Another way of leveraging the neuronal cell culture 100 is by using it directly as a spiking neural network (SNN) [(iii) a trained statistical model]. The flow of data and connections to the electronic computing device will be the same as described above.. Thus, a biological substitute for a SNN can be built simply by using the natural spiking activity of the neurons in the neuronal cell culture 100… [0059] FIG. 4 shows a method 400 that uses a neuronal cell culture to perform a computational task. Method 400 may be performed using the systems and devices shown in FIGS. 1-3... [0062] At operation 406, a computational task is performed by the electronic computing device based on a difference between the input signal and the output signal. The computational task may be any type of computational task conventionally performed by computers [a method for using a biological and artificial neural network (BANN) system to perform a task, … using the BANN system to perform: receiving an input signal to be processed by the BANN in furtherance of performing the task]. The computational task may be a subtask that is part of a larger task involving multiple subtasks. For example, in performing the computational task the neuronal cell culture may function as a reservoir for reservoir computing. The neuronal cell culture may also be used to perform the computational task by performing as a RNN or SNN [a method for using a biological and artificial neural network (BANN) system to perform a task, the BANN system comprising: (i) a multi-electrode array (MEA); (ii) a biological neural network (BNN) comprising neurons arranged on the MEA, (iii) a trained statistical model; and (iv) at least one processor, the method comprising: using the BANN system to perform: receiving an input signal to be processed by the BANN in furtherance of performing the task]. And in [0026] The neuronal cell culture 100 may be used as a compute substrate as described herein without prior training. Development of a neuronal cell culture 100 before use is different from training because the neuronal cell culture 100 is not directed to learning a new behavior while developing. However, it is also possible that the neuronal cell culture 100 will be trained prior to use as a compute substrate... [0047] FIG. 3 illustrates the use of a neuronal cell culture 100 as a reservoir 300 in reservoir computing… The reservoir 300 is typically a random or structured dynamical system that is driven by the input signal and has a large number of internal degrees of freedom. The reservoir 300 acts as a “black box” that transforms the input data into a more complex representation that captures the temporal dynamics of the signal. The output data is then fed into a simple readout mechanism that is trained to map [(iii) a trained statistical model;] the reservoir state to the desired output...)
encoding the input signal to generate at least one stimulation pattern; stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern; measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating; (in [0035] The electronic computing device 102 includes software components that drive the circuitry 110 as well as interpret signals received from the circuitry 110 [encoding the input signal to generate at least one stimulation pattern; stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern]. The software converts input data into instructions that cause the input device 106 to generate a specific input signal to the neuronal cell culture 100. Output signals detected by the output device 108 and provided to the electronic computing device 102 are interpreted by the software and converted into output data. Thus, the software is used to encode information in a way that can be provided to the neuronal cell culture 100 [encoding the input signal to generate at least one stimulation pattern; stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern] and to decode patterns of action potentials [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating] detected by the output device 108. [0036] With this configuration, the neuronal cell culture 100 can be integrated with the electronic computing device 102 creating a computing system that includes both biological and electronic compute substrates. As hardware component in a computing system, the neuronal cell culture 100 may be thought of as a black box that receives input data and generates output data. The behavior and connections of the neurons 104 in the neuronal cell culture 100 map an input to an output. If the responses are predictable or deterministic, they can be mapped back to the stimulus [using the MEA to generate electrical signals in accordance with the at least one stimulation pattern … the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]…; And generating the electric signals as biological neuron spikes, in [0015] Instead of creating approximations of biological neurons in an electronic-computing architecture, this disclosure describes the use of a neuronal cell culture as a compute substrate to process information directly with actual neurons. This compute platform uses a neuronal cell culture as a piece of hardware that functions in conjunction with conventional silicon electronics. Neurons are remarkable among cells in their ability to propagate signals rapidly over large distances. They do this by generating characteristic electrical pulses called action potentials [using the MEA to generate electrical signals in accordance with the at least one stimulation pattern … the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]: voltage spikes that can travel down axons. With each stimuli a neuron builds more cell membrane potential and when the potential reaches activation threshold the neuron fires [using the MEA to generate electrical signals in accordance with the at least one stimulation pattern]. The rate of firing is determined by the rate of stimulation, a natural leak, activation threshold, and a refectory period. The specific connections between individual neurons and methods for processing information within a neuronal cell culture does not need to be elucidated to use a neuronal cell culture as a compute substrate. It is sufficient for the neuronal cell cultures to respond to input signals in a predictable way [using the MEA to generate electrical signals in accordance with the at least one stimulation pattern … the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]...)
generating, based on the measured at least one response, an input for the trained statistical model, wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages; (As depicted in Fig. 3 &4 input is generated for processing a computational task for the claims ANN model, in [0062] At operation 406, a computational task is performed by the electronic computing device based on a difference between the input signal and the output signal [generating, based on the measured at least one response, an input for the trained statistical model]. The computational task may be any type of computational task conventionally performed by computers. The computational task may be a subtask that is part of a larger task involving multiple subtasks [wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages]. For example, in performing the computational task the neuronal cell culture may function as a reservoir for reservoir computing. The neuronal cell culture may also be used to perform the computational task by performing as a RNN or SNN [generating, based on the measured at least one response, an input for the trained statistical model, wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages]. )
processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN; (in [0078] Clause 3. The computing system of clause 1 or 2, wherein the neuronal cell culture comprises differentiated embryonic stem cells or induced pluripotent stem cells… [0081] Clause 6. The computing system of any of clauses 1 to 5, wherein the input device and output device comprise electrodes (202) configured to provide electric signals to the neuronal cell culture as input signals and detect activation potentials of neurons in the neuronal cell culture as output signals [processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN;]. [0082] Clause 7. The computing system of clause 6, wherein the electrodes are configured as a multi-electrode array (MEA) (200). [0083] Clause 8. The computing system of clause 7, wherein the input signal is provided by first electrodes in the MEA contacting the neuronal cell culture at a first location and the output signal is detected by second electrodes in the MEA contacting the neuronal cell culture at a second location… [0085] Clause 10. The computing system of any of clauses 1 to 9, wherein the computing system is configured such that the neuronal cell culture functions as a reservoir (300) in reservoir computing. [0086] Clause 11. The computing system of clause 10, wherein the electronic computing device implements a trainable learning layer (302) that receives an output signal from the neuronal cell culture. [0087] Clause 12. The computing system of clause 11, wherein the neuronal cell culture and the trainable learning layer together are configured to function as a recurrent neural network (RNN) [processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN]. [0088] Clause 13. The computing system of any of clauses 10 to 12, wherein the reservoir is configured to map input signals received from the input device into a different computational space. [0089] Clause 14. The computing system of any of clauses 1 to 8, wherein the computing system is configured such that the neuronal cell culture functions as a spiking neural network (SNN) [processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN]. [0090] Clause 15. The computing system of clause 14, wherein activation potentials of neurons in the neuronal cell culture provides signals for the SNN.)
and using the output from the trained statistical model in furtherance of performing the task. (in [0062] At operation 406, a computational task is performed by the electronic computing device based on a difference between the input signal and the output signal. The computational task may be any type of computational task conventionally performed by computers. The computational task may be a subtask that is part of a larger task involving multiple subtasks. For example, in performing the computational task the neuronal cell culture may function as a reservoir for reservoir computing. The neuronal cell culture may also be used to perform the computational task by performing as a RNN or SNN [and using the output from the trained statistical model in furtherance of performing the task.].)
Yang teaches the claimed trained statical model as the readout layer as depicted in Fig. 3 and noted above.
While one of ordinary skill in the art would understand that the layers used for computations task should trained for performing computational task, Yang does not expressly use the term “trained” for the read-out layer model associated with the biological neural network (BNN) comprising neurons arranged on the multi-electrode array (MEA).
Peter does expressly use the term “trained” for the read-out layer model associated with the biological neural network (BNN) comprising neurons arranged on the multi-electrode array (MEA), as depicted in Figure 4:
PNG
media_image3.png
574
1232
media_image3.png
Greyscale
Figure 4: The proposed architecture for a cyborg using neuronal cultures as a reservoir. The readout layer can be any linear single-layer trained network or a multi-layer network [processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN], in case of more complex or multiple tasks.
And in Pg. 432-434: Recording extracellular activity of neural networks A particular property of neurons is their ability to set up a potential difference between the inside and the outside of their cell membrane. This potential difference is set up by the interplay of diffusion, ion channels and active ion pumps. This cross-membrane potential essentially prepares the neuron for a ’spike’ of electrical activity; a rapid polarity shift (seen as a voltage spike) across the membrane. These spikes may be triggered via electrical or chemical synaptic input from upstream neurons, or by electrical or chemical manipulation of the extracellular environment [training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN]… By growing a dissociated neural network on top of the MEA, it is possible to monitor the extracellular voltage fluctuations that occur in the network in relation to an on-board reference electrode (ground)… Embodiment of Reservoir Computing System We have currently embodied the neuronal culture through a simulated body. Through providing the biological networks sensory feedback in the form of extracellular stimulation, as well as converting network activity into robotic motor behavior, we enable a closed-loop system inspired by the sensory-motor loop vital for animal perception and movement [and using the output from the trained statistical model in furtherance of performing the task]. A simulated robot, or virtual creature (DeMarse et al. (2001) coined the term Animat) provides an easy and safe setup for testing the distributed neuro-robotic system and allows for the environment to be simplified. The animat here, consists of a body with four eyes and two motors, allowing it to sense the environment in a cone and steer either left or right. The environment that the animat inhabits is a small box with no features other than the four enclosing walls. In this very simple environment the animat can be trained to perform simple tasks such as wall avoidance, or skirting the walls without getting too close or too far…
Examiner notes that Peter provides teaches that a biological neural network, BNN, is considered an multi-electrode array network that uses measured extracellular voltage to perform modeling tasks by modeling features as weighted relationships and voltage action pattens based on simulations, as depicted in Fig. 4 and disclosed by the reference in Pg. 432: Recording extracellular activity of neural networks A particular property of neurons is their ability to set up a potential difference between the inside and the outside of their cell membrane. This potential difference is set up by the interplay of diffusion, ion channels and active ion pumps. This cross-membrane potential essentially prepares the neuron for a ’spike’ of electrical activity; a rapid polarity shift (seen as a voltage spike) across the membrane. These spikes may be triggered via electrical or chemical synaptic input from upstream neurons, or by electrical or chemical manipulation of the extracellular environment. During the spike, this polarity change propagates along the neurons axon, a long tendril that can extend to close and far away neurons, causing similar polarity changes, and possibly spikes, in the downstream neurons. It is this electrical property of neurons one can take advantage of in order to record and stimulate a network of neurons by use of an MEA. By growing a dissociated neural network on top of the MEA, it is possible to monitor the extracellular voltage fluctuations that occur in the network in relation to an on-board reference electrode (ground). In addition, one may also stimulate the network by injecting a current through one or several of the MEAs electrodes. This current causes a shift to the cross-membrane potentials of nearby neurons which, if the stimulation is sufficiently strong, may result in these neurons spiking.
Peter and Yang are analogous art because both involve developing systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for using biological neural network as a reservoir of dynamics for modeling and provide control of an embodied agent, as disclosed by Peter with the systems and techniques based on implementing neuronal cell cultures as compute substrates for performing machine learning computations/tasks, as disclosed by Yang.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Peter and Yang to allow for the inclusion of morphogenetic principles into computer architecture to decrease energy consumption and to allow for growing tailored neural networks with target functionalities, (Peter, Pg. 430: Introduction).
Regarding claim 19, the claim limitations are similar with claim 1 and rejected under the same rationale.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Jordan et al. (US 20210334657 hereinafter ‘Jordan’) in view of Li et al. (US 20250094763, hereinafter ‘Li’) and in further view of Peter.
Regarding independent claim 1, Jordan teaches a method for using a biological and artificial neural network (BANN) system to perform a task, the BANN system comprising: (i) a multi-electrode array (MEA); (ii) a biological neural network (BNN) comprising neurons arranged on the MEA, (iii) a trained statistical model; and (iv) at least one processor, the method comprising: (As depicted in Fig. 7, And in [0092] The whole system may thus operate as a Biological Operating System 750 [a method for using a biological and artificial neural network (BANN) system to perform a task], as it gathers all the systems required to ensure the operations of the BNN computing system. [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models [(iii) a trained statistical model], and the functional control software 701 may control those algorithms as well as their parameters in accordance with the end user needs… For instance, pre-processors may be useful to repeat the input application signals over some period of time or to slightly vary them for more robust training and re-enforcement of the learning process [… a task for training and re-enforcement of the learning process].; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) [the BANN system comprising: (i) a multi-electrode array (MEA);]. Alternately, a subset of the neurons may be genetically modified to receive optical stimulation as an input signal from an optogenetic system. The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal…And in [0148] The various operations of example methods described herein may be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors [and (iv) at least one processor, the method comprising] may constitute processor-implemented modules that operate to perform one or more operations or functions described herein. As used herein, processor-implemented module refers to a hardware module implemented using one or more processors. And in 0001: [0001] The present disclosure relates to cognitive computing systems, methods and processes to perform a diversity of high-level, complex cognitive tasks mimicking and extending the biological brain functions…)
using the BANN system to perform: receiving an input signal to be processed by the BANN in furtherance of performing the task; encoding the input signal to generate at least one stimulation pattern; (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface [using the BANN system to perform: receiving an input signal to be processed by the BANN in furtherance of performing the task] such as the electrical signals of a multielectrode array (MEA). Alternately, a subset of the neurons may be genetically modified to receive optical stimulation as an input signal from an optogenetic system. The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [encoding the input signal to generate at least one stimulation pattern]…; And in [0050] In a possible embodiment, the Maxwell MEA micro-sensors may operate as the Stimulation Unit 110. They enable to stimulate from input digital data patterns [encoding the input signal to generate at least one stimulation pattern] the BNN activity using a subset of active stimulation electrode sites. The input digital data patterns 105 may be then be prepared by various data processing methods and software… Each stimulation channel may deliver up to ±1.6 V voltage or ±1.5 mA current amplitude, at an amplitude resolution of 2 nA and a time resolution of 2 s. The MaxLab Live software component may generate various digital data stimulation patterns suitable for these resolutions, such as monophasic, biphasic, triphasic pulses, ramp waveforms, and other custom pulse shapes.)
stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern; (in [0007] Therefore, rather than replicating high-level cognitive processes with AI using silicon based digital computation, an emerging alternate approach consists in using biological neural networks. Recent progress in biotechnology now facilitates the culture and assembly of biological neural networks out of embryonic stem cells such as rat embryonic stem cells, as well as from differentiated human Induced Pluripotent Stem cells (IPSc). The cultured BNN may then be stimulated and read using Multi-Electrode Arrays (MEA) [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern]...; And in [0055] The bio-compatible material may also be specifically adapted to the microelectronic components of the stimulation unit SU 110 and/or the readout unit RU 130... Etching using a mask 252 opposite to the MEA array structure 250 may be used in order to create a biological layer 251 that induces neurons positioning at more precise locations, by aligning the bio-compatible material 251 with the underlying MEA 250 so that the stimulation unit 110 may stimulate the BNN 120 at the neuron level [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one stimulation pattern] and/or the readout unit 130 may read the BNN 120 at the neuron level.
measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating; (in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells... The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern] to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally [stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]… The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure [generating, based on the measured at least one response, an input for the trained statistical model] the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally [wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages;]…; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating;]... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system...; [0124] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135 [to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]. The feedback function (H1) 1137 corresponds to the learning process of the ANN, for example but not limited to a backpropagation. The outer feedback function (H2) 1136 corresponds to a mean to impose long term potentiation to the BNN. The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking at the output of the BNN culture 120 [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern to obtain at least one measured response of the BNN, the at least one measured response comprising a plurality of measured extracellular voltages produced by the BNN responsive to the stimulating]…)
generating, based on the measured at least one response, an input for the trained statistical model, wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages; (in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells... The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern] to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally [wherein the input for the ANN comprises multiple features derived from the plurality of measured extracellular voltages;]… The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure [generating, based on the measured at least one response, an input for the trained statistical model] the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally…; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern; generating, based on the measured at least one response, an input for the trained statistical model]... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one stimulation pattern; generating, based on the measured at least one response, an input for the trained statistical model]. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.)
processing the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model … and using the output from the trained statistical model in furtherance of performing the task. (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA)... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit [processing the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model], using measurement systems such as a multielectrode array (MEA) receiving an electrical signal. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system [and using the output from the trained statistical model in furtherance of performing the task]. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.; And in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms [processing the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model] such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, and the functional control software 701 may control those algorithms as well as their parameters in accordance with the end user needs. Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats. Signal post-processing may comprise transforming the signals output from the BNN [processing the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model] into a more comprehensive format so that they provide the target functionality. In a possible embodiment the BNN may be modeled as a non-linear system and signal pre-processing may comprise applying a non-linear gain to the input signals. In a possible embodiment the BNN functional interface may include one or more artificial neural networks [including the trained statistical model] as pre-processors and/or post-processors [processing the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model … and using the output from the trained statistical model in furtherance of performing the task] and the parameters may comprise weight values, choice of activation functions, and other parameters to achieve the learning of the signals [the input multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model] and/or their classification. For instance, pre-processors may be useful to repeat the input application signals over some period of time or to slightly vary them for more robust training and re-enforcement of the learning process.)
Jordan teaches encoding data as recited electrical signals, as noted above.
Jordan does not expressly teach the training process as claimed in wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN;
Li does expressly teach the training process as claimed in wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN; ([0061-0063: Step 3: Obtain a mapping relationship model between a stimulation sequence and a neural response by training. [0062] Step 3.1: Firstly collect a dataset [wherein the trained statistical model is trained using training data comprising:]: inputting stimulations to the final stimulation input electrode sites and stimulation response electrode sites determined in step 2, and acquiring neural response signals corresponding to the stimulations, i.e., inputting stimulation sequence data to the brain-on-a-chip unit through a stimulation unit [paired inputs and outputs], and meanwhile acquiring and recording real neural response signals of the brain-on-a-chip through a data acquisition unit [paired inputs and outputs]; further processing these neural response signals [with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN] by a data preprocessing unit to extract effective spike sequence data [generated using measured extracellular voltages]; and using the obtained spike sequence data as corresponding neural response data of the stimulation sequence [to application of training stimulation patterns to the BNN]. Step 3.2: Then, train an artificial neural network model using the above dataset [wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN], wherein a single input of the artificial neural network model is a neural response at a certain moment and a stimulation sequence for a period of time, and an output of the artificial neural network model is a predicted neural response for a period of time thereafter; train the model by constructing an objective function to minimize an error between the predicted neural response and a real neural response and adjusting weights in the artificial neural network model, so that the model learns an input-output relationship between the stimulation sequence and the neural response; and finally obtain the mapping relationship model between the stimulation sequence and the neural response…; And in [0009] the data preprocessing unit is configured to process the acquired original neural response signals of the brain-on-a-chip, and extract effective spike signals [generated using measured extracellular voltages from biological cells] as neural response data; And 0002: …. A brain-on-a-chip refers to coupling cortical or hippocampal neurons and a cerebral organ with electrodes [generated using measured extracellular voltages from biological cells] to construct 2D and 3D neurons, i.e., an electronic complex, and to obtain a “brain-on-a-chip” electronic complex with certain intelligence through external neuromodulation and learning training… Therefore, an information interaction platform for the brain-on-a-chip intelligence complex is constructed based on MEAs [generated using measured extracellular voltages from biological cells], which can realize real-time recording and feedback of cell culture [generated using measured extracellular voltages from biological cells] and neural signals. Examiner notes that the spike and neural response signals are considered claimed voltage signals per teaches in Li 00045: Specifically, the brain-on-a-chip basic module includes three constituent units: the brain-on-a-chip, a data acquisition unit and a stimulation unit. The brain-on-a-chip is a coupling body of neurons and an MEAs chip, which, specifically, may be a 2D cortex or hippocampal neuronal network and 3D cerebral organ coupled with the MEAs chip, and reaches maturity after culture. The data acquisition unit is configured to read neural response signals of the brain-on-a-chip in real time, which lays the foundation for further signal decoding. The stimulation unit is configured to apply a corresponding voltage [corresponding the neural signal extracellular voltages] stimulation to the neurons of the brain-on-a-chip according to a stimulation sequence of the brain-on-a-chip information interaction and training module.)
Additionally Li teaches:
and using the output from the trained statistical model in furtherance of performing the task. (in 0065: A correspondence relationship between neural response data (i.e., spike sequence data) and control instructions for the external device [using the output from the trained statistical model in furtherance of performing the task] is constructed based on the neural response data collected in the step 3.1, and the correspondence relationship is embedded into a neural signal decoding unit to enable the neural signal decoding unit to convert…; And depicted in Fig. 2
PNG
media_image4.png
420
772
media_image4.png
Greyscale
And in [0075] FIG. 3 is a schematic diagram of a virtual external device environment of a robot obstacle avoidance task [using the output from the trained statistical model in furtherance of performing the task] designed by this embodiment, which mainly includes a virtual robot (i.e., trolley), a visual workspace for an obstacle avoidance task (i.e., a virtual obstacle scene), a robot status information panel, a learning training and control operation area, as well as stimulation signal setting, data recording and exporting modules…)
Jordan and Li are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for a brain-on-a-chip intelligence complex control system using biological and artificial neural signals and models as disclosed by Li with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as disclosed by Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Li and Jordan in order to produce an accurate neural response according to control needs, (Li, 0022).
Jordan and Yang teaches the claimed system for using electro-arrays to model machine learning systems as claimed where the array produces electrical signals as claimed extracellular voltages produced by the BNN responsive to the stimulating.
While one of ordinary skill in the art would understand that a MEA based BNN produce electrical signals responsive to the stimulating, as claimed extracellular voltages produced by the BNN responsive to the stimulating, as disclosed by the references cited above
Additionally, Peter teaches one of ordinary skill in the art would understand that a MEA based BNN produce electrical signals responsive to the stimulating, as claimed extracellular voltages produced by the BNN responsive to the stimulating and use the term “trained” for the read-out layer model associated with the biological neural network (BNN) comprising neurons arranged on the multi-electrode array (MEA), as depicted in Figure 4:
PNG
media_image3.png
574
1232
media_image3.png
Greyscale
Figure 4: The proposed architecture for a cyborg using neuronal cultures as a reservoir. The readout layer can be any linear single-layer trained network or a multi-layer network [processing the multiple features derived from the plurality of measured extracellular voltages of the at least one response with the trained statistical model to obtain corresponding output from the trained statistical model, wherein the trained statistical model is trained using training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN], in case of more complex or multiple tasks.
And in Pg. 432-434: Recording extracellular activity of neural networks A particular property of neurons is their ability to set up a potential difference between the inside and the outside of their cell membrane. This potential difference is set up by the interplay of diffusion, ion channels and active ion pumps. This cross-membrane potential essentially prepares the neuron for a ’spike’ of electrical activity; a rapid polarity shift (seen as a voltage spike) across the membrane. These spikes may be triggered via electrical or chemical synaptic input from upstream neurons, or by electrical or chemical manipulation of the extracellular environment [training data comprising: paired inputs and outputs with the inputs generated using measured extracellular voltages produced responsive to application of training stimulation patterns to the BNN]… By growing a dissociated neural network on top of the MEA, it is possible to monitor the extracellular voltage fluctuations [measuring generated extracellular voltages] that occur in the network in relation to an on-board reference electrode (ground)… Embodiment of Reservoir Computing System We have currently embodied the neuronal culture through a simulated body. Through providing the biological networks sensory feedback in the form of extracellular stimulation, as well as converting network activity into robotic motor behavior, we enable a closed-loop system inspired by the sensory-motor loop vital for animal perception and movement [and using the output from the trained statistical model in furtherance of performing the task]. A simulated robot, or virtual creature (DeMarse et al. (2001) coined the term Animat) provides an easy and safe setup for testing the distributed neuro-robotic system and allows for the environment to be simplified. The animat here, consists of a body with four eyes and two motors, allowing it to sense the environment in a cone and steer either left or right. The environment that the animat inhabits is a small box with no features other than the four enclosing walls. In this very simple environment the animat can be trained to perform simple tasks such as wall avoidance, or skirting the walls without getting too close or too far…
Examiner notes that Peter provides teaches that a biological neural network, BNN, is considered an multi-electrode array network that uses measured extracellular voltage to perform modeling tasks by modeling features that include weighted relationships, voltage action pattens, and timing properties of volage patterns/spikes based on simulations, as depicted in Fig. 4 and disclosed by the reference in Pg. 432: Recording extracellular activity of neural networks A particular property of neurons is their ability to set up a potential difference between the inside and the outside of their cell membrane. This potential difference is set up by the interplay of diffusion, ion channels and active ion pumps. This cross-membrane potential essentially prepares the neuron for a ’spike’ of electrical activity; a rapid polarity shift (seen as a voltage spike) across the membrane. These spikes may be triggered via electrical or chemical synaptic input from upstream neurons, or by electrical or chemical manipulation of the extracellular environment. During the spike, this polarity change propagates along the neurons axon, a long tendril that can extend to close and far away neurons, causing similar polarity changes, and possibly spikes, in the downstream neurons. It is this electrical property of neurons one can take advantage of in order to record and stimulate a network of neurons by use of an MEA. By growing a dissociated neural network on top of the MEA, it is possible to monitor the extracellular voltage fluctuations that occur in the network in relation to an on-board reference electrode (ground). In addition, one may also stimulate the network by injecting a current through one or several of the MEAs electrodes. This current causes a shift to the cross-membrane potentials of nearby neurons which, if the stimulation is sufficiently strong, may result in these neurons spiking.
Peter, Li and Jordan are analogous art because both involve developing systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for using biological neural network as a reservoir of dynamics for modeling and provide control of an embodied agent, as disclosed by Peter with the systems and techniques based on implementing neuronal cell cultures as compute substrates for performing machine learning computations/tasks, as collectively disclosed by Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Peter, Li and Jordan to allow for the inclusion of morphogenetic principles into computer architecture to decrease energy consumption and to allow for growing tailored neural networks with target functionalities, (Peter, Pg. 430: Introduction).
Regarding claim 2, the rejection of claim 1 is incorporated and Jordan in combination with Li, and Peter further teaches the method of claim 1, wherein the trained statistical model comprises an artificial neural network (ANN). (in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, … Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats…. In a possible embodiment the BNN functional interface may include one or more artificial neural networks [wherein the trained statistical model comprises an artificial neural network (ANN)] as pre-processors and/or post-processors and the parameters may comprise weight values, choice of activation functions, and other parameters to achieve the learning of the signals and/or their classification.; And in [0015] The automation controller may comprise a pre-processing unit to transform, with at least one of a spatio-temporal signal filter, a spatio-temporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computing method, the input data signal into the stimulation signal. [0016] The automation controller may comprise a post-processing unit to transform with at least one of a spatio-temporal signal filter, a spatio-temporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network [wherein the trained statistical model comprises an artificial neural network (ANN)], a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computing method, the readout signal into the output data signal.)
Regarding claim 3, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein encoding the input signal to generate the at least one stimulation pattern comprises encoding the input signal using a second trained statistical model different from the trained statistical model. (in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, … Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats…. In a possible embodiment the BNN functional interface may include one or more artificial neural networks [wherein encoding the input signal to generate the at least one stimulation pattern comprises encoding the input signal using a second trained statistical model different from the trained statistical model in the plurality of networks that can be included] as pre-processors and/or post-processors and the parameters may comprise weight values, choice of activation functions, and other parameters to achieve the learning of the signals [wherein encoding the input signal to generate the at least one stimulation pattern comprises encoding the input signal …] and/or their classification.)
Regarding claim 4, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 3, wherein the second trained statistical model comprises a second ANN. (in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, … Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats…. In a possible embodiment the BNN functional interface may include one or more artificial neural networks [wherein the second trained statistical model comprises a second ANN in the plurality of networks that can be included] as pre-processors and/or post-processors and the parameters may comprise weight values, choice of activation functions, and other parameters to achieve the learning of the signals and/or their classification.)
Regarding claim 5, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein the task is a classification task, a prediction task, a dimensionality reduction task, a reinforcement learning task, or a regression task. (in [0093-0094] … In a possible embodiment the BNN functional interface may include one or more artificial neural networks as pre-processors and/or post-processors and the parameters may comprise weight values, choice of activation functions, and other parameters to achieve the learning of the signals and/or their classification [wherein the task is a classification task]… Typically, the BNN may be trained with a set of k different functions {O(t), S(t)}(k) altogether called the training set where O(t) and S(t) respectively represent the vectors of components Oi(t) and Si(t) respectively. After a successful training, the BNN can be used to predict correct outputs [wherein the task is a …, a prediction task] for inputs S(t) which do not belong to the training set…; And in 0011: …The BNN system may be trained to produce behaviors specified by the reference model through reinforcement learning, for instance by releasing global reward or punishment signals based on behavioral results feedback. Ju Han proposed for instance to use NDMA receptor antagonist drug treatment for reinforcement learning… And in [0117] FIG. 8 illustrates a learning process as may be implemented by the real time processing software, with a feedback loop H 800 to adapt the stimulation signals 605 to the measured readout signals 635… More generally the environmental parameters and the chemical parameters as controlled by the health monitoring system may also be part of a closed loop learning system. In a possible embodiment, the real time processing software may trigger the delivery of a dose of a drug known to reinforce the BNN [a reinforcement learning task] as a reward when the measured readout signals 635 are in line with the training set expected signals for a given stimulation signal input.)
Regarding claim 6, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages; (in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages] on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages] in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes…; [0124] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135 [measuring, using each of the one or more electrodes of the MEA]… The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages], which will strengthen the inner connections inside the BNN…)
and determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold. ( in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold]. The intensity values of the p beams I.sub.p may then be defined such that sum(I.sub.p) T, even if the individual intensity of each beam is lower than the spiking threshold I.sub.p<T [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold where the number of spikes above the zero voltage threshold are considered in the spiking sum]…; And in [0083] In a possible embodiment, the BNN automation controller may directly output the raw BNN readouts signals 635 as the signals resulting from the end-to-end BNN system processing. In an alternate embodiment, the BNN automation controller 600 may further include a post-processing unit 630 in charge with transforming the raw BNN readouts signals 635 into output signals 135. This enables the BNN automation controller to better adapt the raw readouts, which may be too noisy to easily interpret, to the actual application needs. An exemplary application is the extraction of relevant signals out from a train of spikes with a spike sorting signal processing algorithm…)
Additionally, Peter teaches determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold, in Pg. 432: Recording extracellular activity of neural networks A particular property of neurons is their ability to set up a potential difference between the inside and the outside of their cell membrane. This potential difference is set up by the interplay of diffusion, ion channels and active ion pumps. This cross-membrane potential essentially prepares the neuron for a ’spike’ of electrical activity; a rapid polarity shift (seen as a voltage spike) across the membrane. These spikes may be triggered via electrical [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold above zero] or chemical synaptic input from upstream neurons, or by electrical or chemical manipulation of the extracellular environment. During the spike, this polarity change propagates along the neurons axon, a long tendril that can extend to close and far away neurons, causing similar polarity changes, and possibly spikes, in the downstream neurons. It is this electrical property of neurons one can take advantage of in order to record and stimulate a network of neurons by use of an MEA. By growing a dissociated neural network on top of the MEA, it is possible to monitor the extracellular voltage fluctuations that occur in the network in relation to an on-board reference electrode (ground) [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold above reference electrode]…
It would have been obvious to one of ordinary skill in the art before the effective filing date of the present application to combine the teachings of Peter, Li and Jordan for the same reasons disclosed above.
Regarding claim 7, the rejection of claim 6 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 6, wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating. ([0126] Then H2 is a function or an algorithm (found by example with the help of genetic programming or machine learning) that minimizes (or maximizes) a scalar metric L so that when L is minimal (or maximal) for an arbitrary number of periods P, the BNN produces always the same spiking temporal function O(t) for a given input S(t) [wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating]. L is known as a loss (or reward) function.)
Regarding claim 8, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein measuring, using the MEA, the at least one response of the BNN comprises measuring spectral information. (in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 [wherein measuring, using the MEA, the at least one response of the BNN comprises measuring spectral information] may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes. The intensity values of the p beams I.sub.p may then be defined such that sum(I.sub.p) T, even if the individual intensity of each beam is lower than the spiking threshold I.sub.p<T. As will be apparent to those skilled in the art of optogenetics, various lighting means [wherein measuring, using the MEA, the at least one response of the BNN comprises measuring spectral information.] may be used in accordance with the genetically modified photosensitive neural cells characteristics. In a possible embodiment, lasers for optogenetics may be used as the lighting beams. In order to excite a specific neuron lying at a given depth inside a 3D agglomerate of neurons, a laser (or multispectral) beam [wherein measuring, using the MEA, the at least one response of the BNN comprises measuring spectral information.] may be focused in order to reach the maximum energy of light exactly at the neuron depth location. The targeted position can further be adjusted by either changing the location of the light emitting device and/or by adjusting orientation of mirrors used to deflect the beam direction.; And in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server… The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally. Examples of realizations of the SU 110 include but are not limited to multi-electrodes arrays (MEA), patch-clamps, light induced stimulation such as optogenetics systems [wherein measuring, using the MEA, the at least one response of the BNN comprises measuring spectral information], magnetic or electric fields, ion stimulation, focused laser light, optical tweezers, or mechanically induced stimuli through gravity or pressure changes. The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally. Examples of realizations of the RU 130 include but are not limited to multi-electrodes arrays (MEA), patch-clamps, imaging systems, ion sensitive sensors, electrical or magnetic sensitive sensors, chemical sensors, and other sensors suitable to neuron cultures.)
Regarding claim 9, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein measuring, using the MEA, the at least one response of the BNN comprises deriving from the at least one response of the BNN, multiple features. (in [0014] Automated processing systems are described for transforming a spatio-temporal input data signal into a spatio-temporal output data signal [wherein measuring, using the MEA, the at least one response of the BNN comprises deriving from the at least one response of the BNN, multiple features], … BNN culture, one or more sensors to measure at least one environmental parameter of the BNN culture, and an automation controller configured to adapt the stimulation signal to the input data signal, to adapt the output data signal to the readout signal and to control at least one of a BNN core unit environmental parameter, [wherein measuring, using the MEA, the at least one response of the BNN comprises deriving from the at least one response of the BNN, multiple features] a BNN neural cell culture nutrient supply, a BNN neural cell culture additive supply, a BNN neural cell culture nutrient waste collection, a BNN neural cell culture additive waste collection so as to maintain the homeostasis of the BNN neural cell culture over time such that the spatio-temporal input data signal is continuously transformed into the spatio-temporal output data signal. And in [0084] ….The proposed BOS operates with the BNN core unit culture 120, the stimulation module and readout module SU/RU 110/130, and a BNN health control unit 710 for an automated vascularization system (AVS) as formerly described. The BNN health control unit 710 may comprise a chemical control unit with the nutrient and additive tanks and dispensers as well as the waste collectors, as well as an environmental control unit in charge with controlling other BNN culture environment parameters [wherein measuring, using the MEA, the at least one response of the BNN comprises deriving from the at least one response of the BNN, multiple features] such as temperature, pressure, humidity, ratio of O2 or CO2, and other parameters. It may include for instance digitally controlled micro-pumps for the delivery of chemicals as well as temperature, hygrometry, pH or CO2 sensors, or it may even change any other environmental parameters including for instance sound or light waves in any frequency band. The BNN health control unit is in charge with monitoring the health of the BNN cells under real time supervision by a BNN health control software 700. This health monitoring system thus operates as a homeostatic system which in charge of regulating the BNN performance through time, as it may naturally drift over time, transparently to the end user. It operates by adjusting the environmental parameters (chemicals, nutrients, temperature, CO2, etc.) to ensure proper functionality of the BNN processing unit over time. This may include periodically checking that the SU input signals training set still results in the expected RU output signals. The BNN health control software 700 may be managed by a user administrator of the system through an administrator interface 740. )
Regarding claim 10, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 9, wherein the multiple features comprises one or more of spike rate, latency, average latency, a sequence of images of the at least one response of the BNN, and/or earth mover's distance. ([0101] Preferably, the pre-processing unit may perform a spatio-temporal processing of the raw inputs Sr(t) in order to generate the processed signals S(t) to be fed into the BNN with the SU. The BNN then outputs raw signals O.sub.r(t) which are in turn processed by the post-processing unit in order to generate the processed output signals O(t). The dimensionality of each of these signals may be different. For instance, for the BNN to output one spike for a simple raw one-dimensional sinus signal of 1 hz frequency [wherein the multiple features comprises one or more of spike rate, latency…], it may be required to excite 100 different electrodes one after the other, each with a different temporal signal defined at a resolution of 100 hz [wherein the multiple features comprises one or more of spike rate, latency…]…)
Regarding claim 11, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising: prior to stimulating the BNN, selecting a subset of a plurality of electrodes of the MEA to use when stimulating the BNN, and wherein the stimulating the BNN using the MEA comprises stimulating the BNN using only the selected subset of the plurality of electrodes. (in [0039] A subset of the neurons [further comprising: prior to stimulating the BNN, selecting a subset of a plurality of electrodes of the MEA to use when stimulating the BNN] may be excited with an input signal through a BNN [and wherein the stimulating the BNN using the MEA comprises stimulating the BNN using only the selected subset of the plurality of electrodes.] input interface such as the electrical signals of a multielectrode array (MEA) [further comprising: prior to stimulating the BNN, selecting a subset of a plurality of electrodes of the MEA to use when stimulating the BNN]. Alternately, a subset of the neurons may be genetically modified to receive optical stimulation as an input signal from an optogenetic system. The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal…)
Regarding claim 12, the rejection of claim 11 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 11, wherein selecting the subset of the plurality of electrodes comprises: stimulating the BNN by using the plurality of electrodes of the MEA to generate electrical signals in accordance with at least one calibration stimulation pattern; measuring, using the MEA, at least one response of the BNN to being stimulated with the at least one calibration stimulation pattern; (in [0050] In a possible embodiment, the Maxwell MEA micro-sensors may operate as the Stimulation Unit 110. They enable to stimulate from input digital data patterns [wherein selecting the subset of the plurality of electrodes comprises: stimulating the BNN by using the plurality of electrodes of the MEA to generate electrical signals in accordance with at least one calibration stimulation pattern] the BNN activity using a subset of active stimulation electrode sites. The input digital data patterns 105 may be then be prepared by various data processing methods and software. In a possible embodiment the Maxwell Stimulation Module may be used as the SU 110 to provide 32 stimulation channels. Each stimulation channel may deliver up to ±1.6 V voltage or ±1.5 mA current amplitude, at an amplitude resolution of 2 nA and a time resolution of 2 s. The MaxLab Live software component may generate various digital data stimulation patterns suitable for these resolutions, such as monophasic, biphasic, triphasic pulses, ramp waveforms, and other custom pulse shapes.)
and selecting, based on the measured at least one response of the BNN, the subset of the plurality of electrodes based on an amount of neuronal activity induced by the respective ones of the plurality of electrodes. (in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells... The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer [and selecting, based on the measured at least one response of the BNN, the subset of the plurality of electrodes based on an amount of neuronal activity induced by the respective ones of the plurality of electrodes] to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally… The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally…; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) ... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [and selecting, based on the measured at least one response of the BNN, the subset of the plurality of electrodes based on an amount of neuronal activity induced by the respective ones of the plurality of electrodes]. Alternately, a subset of the neurons may be genetically modified to express fluorescence [based on an amount of neuronal activity induced by the respective ones of the plurality of electrodes] as an output signal from the BNN to an imaging sensor system [and selecting, based on the measured at least one response of the BNN, the subset of the plurality of electrodes based on an amount of neuronal activity induced by the respective ones of the plurality of electrodes]. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.)
Regarding claim 13, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 12, wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes; and selecting the subset of the plurality of electrodes based on the ranking. (in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes [wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes]. The intensity values of the p beams I.sub.p [and selecting the subset of the plurality of electrodes based on the ranking] may then be defined such that sum(I.sub.p) T, even if the individual intensity of each beam is lower than the spiking threshold I.sub.p<T. As will be apparent to those skilled in the art of optogenetics, various lighting means may be used in accordance with the genetically modified photosensitive neural cells characteristics…)
Regarding claim 14, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising: subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN. (in [0124-0126] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135. The feedback function (H1) 1137 corresponds to the learning process of the ANN, for example but not limited to a backpropagation. The outer feedback function (H2) 1136 corresponds to a mean to impose long term potentiation to the BNN. The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking [subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN] at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence, which will strengthen the inner connections inside the BNN. Using this approach, the ANN may even be omitted in some embodiments… Then H2 is a function or an algorithm (found by example with the help of genetic programming or machine learning) that minimizes (or maximizes) a scalar metric L so that when L is minimal (or maximal) for an arbitrary number of periods P [subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN], the BNN produces always the same spiking temporal function O(t) for a given input S(t). L is known as a loss (or reward) function….)
Regarding claim 15, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein the BANN system further comprises a graphical user interface (GUI) for receiving user input, the user input comprising one or more values for one or more parameters of the at least one stimulation pattern. (in [0086] An end user may interact with the real time control software 701 by means of a user interface 741. The user interface 741 may be the same or different user interface for an end user and/or and an administrator user, but if they are the same, the administrator user has access to more functionality. Parameters for the real-time functional control software 701 [wherein the BANN system further comprises a graphical user interface (GUI) for receiving user input, the user input comprising one or more values for one or more parameters of the at least one stimulation pattern.] as well as for the BNN health control software and/or the BNN functional interface may be stored in a database 720. [0087] For high level cognitive BNN processing unit functionality possibly requiring very low latency and high data throughput bandwidth, the BNN functional interface 730 may apply further data pre-processing 610 and/or data post-processing algorithms 630 under control by the real time functional control software 701. In a possible embodiment, the real-time control software does not operate over the internet but in local. The functional control software 701 may thus be uploaded either by the administrator using the administrator interface 740, or directly by the end user using the user interface 741 [wherein the BANN system further comprises a graphical user interface (GUI) for receiving user input, the user input comprising one or more values for one or more parameters of the at least one stimulation pattern.].)
Regarding claim 16, the rejection of claim 2 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 2, wherein the ANN comprises a neural network having one or more convolutional layer or a neural network having a transformer architecture. (in [0095] In some embodiments, in order to achieve this successful training, machine learning algorithms (artificial neural networks ANN, convolutional neural networks CNN [wherein the ANN comprises a neural network having one or more convolutional layer], support vector machines SVM and deep learning in particular, but this includes also any ML approaches like random forest, genetic algorithms, genetic programming, reservoir computing,…)
Regarding claim 17, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising: determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern. (in [0124] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135. The feedback function (H1) 1137 corresponds to the learning process of the ANN [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern], for example but not limited to a backpropagation. The outer feedback function (H2) 1136 corresponds to a mean to impose long term potentiation to the BNN. The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern], which will strengthen the inner connections inside the BNN. Using this approach, the ANN may even be omitted in some embodiments.)
Regarding claim 18, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising optimizing the biological neural network to perform the task. (in [0014] Automated processing systems are described for transforming a spatio-temporal input data signal into a spatio-temporal output data signal, the system comprising: an in vitro biological culture of neural cells (BNN core unit) [comprising optimizing the biological neural network to perform the task], an input stimulation unit (SU) adapted to apply an input spatio-temporal stimulation signal into a first set of the neural cells, an output readout unit (RU) adapted to capture an output spatio-temporal readout signal from a second set of the neural cells, one or more nutrient tanks in connection with one or more nutrient dispensers to inject one or more nutrients into the biological neural cell culture, one or more additive tanks each in connection with one or more additive dispensers to inject one or more additives into the BNN culture, one or more nutrient waste collectors for filtering and expelling nutrient waste from the BNN culture, one or more additive waste collectors for filtering and expelling additive waste from the BNN culture, one of more vascularization networks to connect said nutrient dispensers, additive dispensers, nutrient waste collectors and additive waste collectors to the BNN culture, one or more sensors to measure at least one environmental parameter of the BNN culture, and an automation controller configured to adapt the stimulation signal to the input data signal, to adapt the output data signal to the readout signal and to control at least one of a BNN core unit environmental parameter, a BNN neural cell culture nutrient supply, a BNN neural cell culture additive supply, a BNN neural cell culture nutrient waste collection, a BNN neural cell culture additive waste collection so as to maintain the homeostasis of the BNN neural cell culture over time [comprising optimizing the biological neural network to perform the task] such that the spatio-temporal input data signal is continuously transformed into the spatio-temporal output data signal. And in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells [comprising optimizing the biological neural network to perform the task]. The cells may be assembled through a variety of different processes, for instance, but not limited to, cell cultures or organogenesis like approaches. Throughout this disclosure, the terminology cell culture is used indifferently for in vitro cells growth and lifetime maintenance out of their natural in vivo environment. The BNN cells may be arranged in 2D or in 3D [comprising optimizing the biological neural network to perform the task]….)
Regarding claim 19, the claim limitations are similar to the claim limitations in claim 1, and the claims are rejected under the same rationale.
Regarding independent claim 20, Jordan teaches encoding the input signal to generate at least one stimulation pattern; (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA). Alternately, a subset of the neurons may be genetically modified to receive optical stimulation as an input signal from an optogenetic system. The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [encoding the input signal using the trained statistical model to generate at least one temporal sequence of two-dimensional stimulation patterns]…; And in [0050] In a possible embodiment, the Maxwell MEA micro-sensors may operate as the Stimulation Unit 110. They enable to stimulate from input digital data patterns [encoding the input signal using the trained statistical model to generate at least one temporal sequence of two-dimensional stimulation patterns;] the BNN activity using a subset of active stimulation electrode sites. The input digital data patterns 105 may be then be prepared by various data processing methods and software… Each stimulation channel may deliver up to ±1.6 V voltage or ±1.5 mA current amplitude, at an amplitude resolution of 2 nA and a time resolution of 2 s [at least one temporal sequence of two-dimensional stimulation patterns as a spike signal having a two dimensions amplitude and time pules/waveforms]. The MaxLab Live software component may generate various digital data stimulation patterns suitable for these resolutions, such as monophasic, biphasic, triphasic pulses, ramp waveforms, and other custom pulse shapes….)
stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns; (in [0007] Therefore, rather than replicating high-level cognitive processes with AI using silicon based digital computation, an emerging alternate approach consists in using biological neural networks. Recent progress in biotechnology now facilitates the culture and assembly of biological neural networks out of embryonic stem cells such as rat embryonic stem cells, as well as from differentiated human Induced Pluripotent Stem cells (IPSc) [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns;]. The cultured BNN may then be stimulated and read using Multi-Electrode Arrays (MEA) [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns]...; And in [0055] The bio-compatible material may also be specifically adapted to the microelectronic components of the stimulation unit SU 110 and/or the readout unit RU 130... Etching using a mask 252 opposite to the MEA array structure 250 may be used in order to create a biological layer 251 that induces neurons positioning at more precise locations, by aligning the bio-compatible material 251 with the underlying MEA 250 so that the stimulation unit 110 may stimulate the BNN 120 at the neuron level [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns] and/or the readout unit 130 may read the BNN 120 at the neuron level.)
measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns; (in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells... The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns] to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally… The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally…; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns]... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns]. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.)
and using the measured at least one response from the BNN in furtherance of performing the task. (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA)... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns], using measurement systems such as a multielectrode array (MEA) receiving an electrical signal. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system [and using the measured at least one response from the BNN in furtherance of performing the task]. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.; And in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, and the functional control software 701 may control those algorithms as well as their parameters in accordance with the end user needs. Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats. Signal post-processing may comprise transforming the signals output from the BNN to a more comprehensive format so that they provide the target functionality. In a possible embodiment the BNN may be modeled as a non-linear system and signal pre-processing may comprise applying a non-linear gain to the input signals...)
Additionally, Peter teaches … at least one temporal sequence of two-dimensional stimulation patterns; as depicted in Figure :
And in Pg. 431: Reservoir Computing Artificial neural networks (ANNs) represent a class of computational models that take inspiration from biological neural networks (BNNs). In ANNs, artificial neurons (the computing elements) are typically arranged in layers, i.e. neurons in one layer are connected to neurons in the next layer and information flows in a feed-forward fashion. Artificial recurrent neural networks (RNNs) are a more plausible and realistic model of BNNs, where the connection topology of neurons has recurrencies, i.e. cycles [… at least one temporal sequence of two-dimensional stimulation patterns]. By allowing cycles, the RNN becomes a dynamic system with a self-sustained temporal activation and possesses memory of previous inputs, i.e. the activation state of the network is a function of previous activation states. Such property is known as ”echo state” (Jaeger, 2003). Artificial RNNs are far more powerful than feed-forward ANNs but they are also far more difficult to train, because learning gradients dissipate over time (making it difficult to learn long-range memory dependencies) and network dynamics can lead to bifurcations. It has recently been suggested that both artificial and biological RNNs may be considered as a high-dimensional medium of dynamics with the ability to represent information in a high-dimensional and discriminating space [… at least one temporal sequence of two-dimensional stimulation patterns]…
Peter, Li and Jordan are analogous art because both involve developing systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for using biological neural network as a reservoir of dynamics for modeling and provide control of an embodied agent, as disclosed by Peter with the systems and techniques based on implementing neuronal cell cultures as compute substrates for performing machine learning computations/tasks, as collectively disclosed by Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Peter, Li and Jordan to allow for the inclusion of morphogenetic principles into computer architecture to decrease energy consumption and to allow for growing tailored neural networks with target functionalities, (Peter, Pg. 430: Introduction).
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Jordan et al. (US 20210334657 hereinafter ‘Jordan’) in view of Li and Peter and in further view of Howard Newton (US 20220111212, hereinafter ‘New’).
Regarding claim 6, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages; (in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages] on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages] in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes…; [0124] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135 [measuring, using each of the one or more electrodes of the MEA]… The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence [wherein measuring, using the MEA, the at least one response of the BNN comprises: measuring, using each of the one or more electrodes of the MEA, a respective series of one or more voltages], which will strengthen the inner connections inside the BNN…)
and determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold. ( in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold]. The intensity values of the p beams I.sub.p may then be defined such that sum(I.sub.p) T, even if the individual intensity of each beam is lower than the spiking threshold I.sub.p<T [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold]…; And in [0083] In a possible embodiment, the BNN automation controller may directly output the raw BNN readouts signals 635 as the signals resulting from the end-to-end BNN system processing. In an alternate embodiment, the BNN automation controller 600 may further include a post-processing unit 630 in charge with transforming the raw BNN readouts signals 635 into output signals 135. This enables the BNN automation controller to better adapt the raw readouts, which may be too noisy to easily interpret, to the actual application needs. An exemplary application is the extraction of relevant signals out from a train of spikes with a spike sorting signal processing algorithm…)
Additionally, New teaches and determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold. (in [0352] Electrophysiological Recording. In electrophysiology—the oldest strategy for neural recording, an electrode is used to measure the local voltage at a recording site, which conveys information about the spiking activity of one or more nearby neurons. The number of recording sites may be smaller than the number of neurons recorded [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold] since each recording site may detect signals from multiple neurons in the area. And in [0437] Spike Sorting. An exemplary data flow block diagram of a spike sorting technique 3900 is shown in FIG. 39. As shown in this example, when data arrives in a data buffer 3902, spike detection 3904 may be performed, using, for example, an adaptive threshold 3906 to recognize spiking events [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold], template memory 3908 to identify neurons, and correlation detector 3910 to identify overlapping spikes.; And in [0450] Electrical Recording Data. In embodiments, the methods described above may be used when the goal is to preserve the waveforms of the spikes. For a higher compression ratio, but at the cost of losing raw waveform information, embodiments may use spike detection and/or spike sorting [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold]. Examples of hardware implementations of spike detection may be as simple as a comparator with a pre-defined threshold [determining a number of spikes measured based on a number of measured voltages exceeding a voltage threshold]. In this way, a compression ratio higher than 100× may be achieved with little power consumption.)
New, Peter, Li and Jordan are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for developing an adaptive threshold to recognize spiking events, template memory to identify neurons in biological artificial neural networks as disclosed by New with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Kag, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Kag, Li and Jordan in order to provide the capability to provide brain monitoring and stimulation devices using parameter sets that are automatically refined during operation,(New, Abstract).
Claims 7 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Jordan et al. (US 20210334657 hereinafter ‘Jordan’) in view Li, in further view of Peter and Davila et al. (US 20190249147, hereinafter ‘Dav’).
Regarding claim 7, the rejection of claim 6 is incorporated and Jordan in combination with Li and Kag further teaches the method of claim 6, wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating. ([0126] Then H2 is a function or an algorithm (found by example with the help of genetic programming or machine learning) that minimizes (or maximizes) a scalar metric L so that when L is minimal (or maximal) for an arbitrary number of periods P, the BNN produces always the same spiking temporal function O(t) for a given input S(t) [wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating]. L is known as a loss (or reward) function.)
Additionally, Dav expressly teaches spike temporal activity are known to have the claimed characteristics in the limitation wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating, in [0053] Neuronal activity parameters include, without limitation, total number of spikes (per recording period) [wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating]; mean firing rate (of spikes) [wherein measuring, using the MEA, the at least one response of the BNN comprises determining an average number of spikes per time period in a plurality of time periods following the stimulating]; inter-spike interval (distance between sequential spikes); total number of bursts (per recording period); burst frequency; number of spikes per burst; burst duration (in milliseconds); inter-burst interval (distance between sequential bursts); burst percentage (the portion of spikes occurring within a burst); total number of network bursts (spontaneous synchronized network activity); network burst frequency; number of spikes per network burst; network burst duration; inter-network-burst interval; inter-spike interval within network bursts; network burst percentage (the portion of bursts occurring within a network burst); cross-correlation of detected spikes between all electrodes per well (e.g. for MEA recordings, measure of synchrony, see FIG. 2B).
Dav, Peter, Li and Jordan are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing compositions and methods for biologically relevant screening of neural function, including determination of the effects of an agent on neural cells using artificial neural network models as disclosed by Dav with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Peter, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Dav, Peter, Li and Jordan in order to implement functional neural networks capable of spontaneous synchronous firing that can be used for phenotypic screening and other purposes, (Dav, 0013 & Abstract).
Regarding claim 14, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising: subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN. (in [0124-0126] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135. The feedback function (H1) 1137 corresponds to the learning process of the ANN, for example but not limited to a backpropagation. The outer feedback function (H2) 1136 corresponds to a mean to impose long term potentiation to the BNN. The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking [subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN] at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence, which will strengthen the inner connections inside the BNN. Using this approach, the ANN may even be omitted in some embodiments… Then H2 is a function or an algorithm (found by example with the help of genetic programming or machine learning) that minimizes (or maximizes) a scalar metric L so that when L is minimal (or maximal) for an arbitrary number of periods P [subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN], the BNN produces always the same spiking temporal function O(t) for a given input S(t). L is known as a loss (or reward) function….)
Additionally, Dav teaches subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN. (in [0023] FIG. 8A-8B. Synchronized network activity in neural co-cultures containing either primary glial cells derived from mice or human glial cells differentiated from early glial progenitors. (FIG. 8A) Raster plots showing different frequencies of synchronized network bursts (same time scale) and reduced firing between bursts [subsequent to measuring the at least one response of the BNN and using the BANN system, stimulating the BNN by using the MEA to generate electrical signals in accordance with at least one calibration pattern designed for reducing burstiness of the BNN] in co-cultures using human versus mouse glial cells. (FIG. 8B) Patch clamp analysis measuring excitatory postsynaptic currents (EPSCs) of single neurons from co-cultures using human glial cells (left) or mouse glial cells (right).; And in [0034] In the context of an MEA system, detection of action potentials in neural cultures, signals can be detected as spikes when exceeding a present voltage increase, e.g. 2×, 3×, 4×, 5×, 6× or more the standard deviation of average voltages measured by each electrode. A set of sequential spikes may be defined as a burst if at least about 3, about 4, about 5 or more spikes are detected by one electrode within a defined period of time, e.g. from around about 10-500 milliseconds, around about 50 to about 250 milliseconds, or around about 100 milliseconds. Bursts detected across multiple electrodes per well can be defined as synchronized network bursts if the first spikes of individual bursts are co-occurring…)
Dav, Peter Li and Jordan are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing compositions and methods for biologically relevant screening of neural function, including determination of the effects of an agent on neural cells using artificial neural network models as disclosed by Dav with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Peter, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Dav, Li and Jordan in order to implement functional neural networks capable of spontaneous synchronous firing that can be used for phenotypic screening and other purposes, (Dav, 0013 & Abstract).
Claims 13, 17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Jordan et al. (US 20210334657 hereinafter ‘Jordan’) in view Li, Peter, and Kagan et al. (US 20250087338, hereinafter ‘Kag’).
Regarding claim 13, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 12, wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes; and selecting the subset of the plurality of electrodes based on the ranking. (in [0058] FIG. 2d) illustrates a possible embodiment of an optogenetics stimulation unit SU interface for generating spikes on genetically modified photosensitive neurons 293. Several light beams 290, 291 and 292 may be targeted on the same neuron, in a way that all beams cross at the location of the neural cell 293, which in turn generates a spike. This method enables to generate spikes in 3D conglomerates of neural cells. Let T, be the intensity threshold of the light stimulation above which an illuminated neuron spikes [wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes]. The intensity values of the p beams I.sub.p [and selecting the subset of the plurality of electrodes based on the ranking] may then be defined such that sum(I.sub.p) T, even if the individual intensity of each beam is lower than the spiking threshold I.sub.p<T. As will be apparent to those skilled in the art of optogenetics, various lighting means may be used in accordance with the genetically modified photosensitive neural cells characteristics…)
Additionally, Kag teaches wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes; and selecting the subset of the plurality of electrodes based on the ranking. (in [0170] ... Accordingly, in some embodiments all of the possible configurations to be tested are determined a priori prior running the online optimization model. In other embodiments, no configurations, or a small sample of starting configurations, may be provided to the online optimization model, and the online optimization model may generate multiple different configurations to test. The scores may be continually updated as further tests are run on the various configurations. Ultimately, the configuration having the highest score may be selected for a neural culture [determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes; and selecting the subset of the plurality of electrodes based on the ranking].; And in [0178] In embodiments, the neural culture may not be perfectly symmetrical. For example, there may be more neurons on one side of the MEA than on another side of the MEA. Additionally, regardless of the number of neurons on different regions of the MEA, it may be technically difficult to culture neurons that display perfectly symmetrical electrical activity in different regions (e.g., in the different motor regions shown in FIG. 7). As a result, some motor regions (or other output regions) may exhibit a stronger electrical response than other motor regions (or other output regions) and/or have a higher baseline electrical activity than other motor regions (or other output regions) [wherein the selecting the subset of the plurality of electrodes comprises: determining, based on the measured at least one response of the BNN, a ranking of respective ones of the plurality of electrodes based on the amount of neuronal activity induced by the respective ones of the plurality of electrodes; and selecting the subset of the plurality of electrodes based on the ranking]. This may bias the system to detect some types of outputs (e.g., move paddle left) more than other types of outputs (e.g., move paddle right).)
Jordan, Li, Peter and Kag are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for interfacing a computing device with biological neurons as disclosed by Kag with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Peter, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Kag, Peter, Li and Jordan in order to provide neural systems that can self-organize to behave in a way which maximizes the complexity of information input and adapts the system’s behavior to increase performance in a task over time, (Kag, 0027).
Regarding claim 17, the rejection of claim 1 is incorporated and Jordan in combination with Li and Peter further teaches the method of claim 1, further comprising: determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern. (in [0124] FIG. 11 illustrates a specific realization related to FIG. 10b) where the O output signal post-processing block 1010 is realized with an artificial neural network (ANN). The Biological Neural Network (BNN) 120 is connected to the digital interfaces Readout Unit (RU) and Stimulation Unit (SU) by means of a Multi-electrodes Array (MEA) 1135. The feedback function (H1) 1137 corresponds to the learning process of the ANN [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern], for example but not limited to a backpropagation. The outer feedback function (H2) 1136 corresponds to a mean to impose long term potentiation to the BNN. The outer feedback function H2 may be achieved with genetic programming or machine learning, so as to determine the correct spatiotemporal sequence that will make the desired spiking at the output of the BNN culture 120. Another way to impose long term potentiation is to use RU as a stimulation unit with the desired spiking sequence [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern], which will strengthen the inner connections inside the BNN. Using this approach, the ANN may even be omitted in some embodiments.)
Kag teaches determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern. (in [0027] …. In one example, the neural system will self-organize to behave in a way that ensures continued stimuli (e.g., that avoids situations in which stimuli is withheld). In one example, the neural system will self-organize to behave in a way that maximizes positive feedback [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern] and minimizes negative feedback. In one example, the neural system will self-organize to behave in a way which maximizes the complexity of information input. ; And in [0080] To train a neural culture to perform tasks (i.e., to play Pong or perform other tasks), an area of an MEA (e.g., grid of electrodes) or other device (e.g., a fully optical device) may be divided into regions, and each region may be assigned a role. An example is set forth below with reference to FIG. 7. One role is a simulated sensory area, in which inputs associated with a virtual environment are provided to the neural culture. The simulated sensory area may also be used for feedback (e.g., punishment stimulus and/or reward stimulus [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern]) during training of the neural culture. Alternatively a role of feedback may be assigned to another region….; And in [0154] ... In this example, the biological neural network may be trained to move the paddle to intercept the ball. Electrophysiological activity of pre-defined motor regions may be recorded to determine how the ‘paddle’ would move, for example. This may be achieved by demarcating the 2D grid in the MEA 105 into 4 quadrants. With each set of electrical/optical impulses that are applied to the biological neural network, the electrical signals generated by the neurons may be measured. If a majority of electrical signals measured are from an upper right quadrant, then this may cause the virtual environment to move the right paddle up. If a majority of electrical signals measured are from a lower right quadrant, then this may cause the virtual environment to move the right paddle down. A positive reward stimulus [determining, based on the measured at least one response of the BNN, whether to apply a positive feedback stimulation pattern to the BNN; and stimulating, based on the determining whether to apply the positive feedback stimulation pattern to the BNN, the BNN with the positive feedback stimulation pattern] or other feedback or lack of feedback may then be provided to the biological neural network when the ball intercepts the right paddle, as discussed above.)
Jordan, Li, Peter and Kag are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for interfacing a computing device with biological neurons as disclosed by Kag with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Peter, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Kag, Peter, Li and Jordan in order to provide neural systems that can self-organize to behave in a way which maximizes the complexity of information input and adapts the system’s behavior to increase performance in a task over time, (Kag, 0027).
Regarding independent claim 20, Jordan teaches encoding the input signal to generate at least one stimulation pattern; (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA). Alternately, a subset of the neurons may be genetically modified to receive optical stimulation as an input signal from an optogenetic system. The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [encoding the input signal using the trained statistical model to generate at least one temporal sequence of two-dimensional stimulation patterns]…; And in [0050] In a possible embodiment, the Maxwell MEA micro-sensors may operate as the Stimulation Unit 110. They enable to stimulate from input digital data patterns [encoding the input signal using the trained statistical model to generate at least one temporal sequence of two-dimensional stimulation patterns;] the BNN activity using a subset of active stimulation electrode sites. The input digital data patterns 105 may be then be prepared by various data processing methods and software… Each stimulation channel may deliver up to ±1.6 V voltage or ±1.5 mA current amplitude, at an amplitude resolution of 2 nA and a time resolution of 2 s [at least one temporal sequence of two-dimensional stimulation patterns as a spike signal having a two dimensions amplitude and time pules/waveforms]. The MaxLab Live software component may generate various digital data stimulation patterns suitable for these resolutions, such as monophasic, biphasic, triphasic pulses, ramp waveforms, and other custom pulse shapes….)
stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns; (in [0007] Therefore, rather than replicating high-level cognitive processes with AI using silicon based digital computation, an emerging alternate approach consists in using biological neural networks. Recent progress in biotechnology now facilitates the culture and assembly of biological neural networks out of embryonic stem cells such as rat embryonic stem cells, as well as from differentiated human Induced Pluripotent Stem cells (IPSc) [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns;]. The cultured BNN may then be stimulated and read using Multi-Electrode Arrays (MEA) [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns]...; And in [0055] The bio-compatible material may also be specifically adapted to the microelectronic components of the stimulation unit SU 110 and/or the readout unit RU 130... Etching using a mask 252 opposite to the MEA array structure 250 may be used in order to create a biological layer 251 that induces neurons positioning at more precise locations, by aligning the bio-compatible material 251 with the underlying MEA 250 so that the stimulation unit 110 may stimulate the BNN 120 at the neuron level [stimulating the BNN by using the MEA to generate electrical signals in accordance with the at least one temporal sequence of two-dimensional stimulation patterns] and/or the readout unit 130 may read the BNN 120 at the neuron level.)
measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns; (in [0038] FIG. 1 depicts the core BNN unit 100 as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but is not limited to, an active biological culture 120, typically an assembly of a plurality of living neural and glial cells... The BNN cells may be arranged in 2D or in 3D. The Stimulation Unit 110 (SU) represents the input interface between a digital data input signal 105 and the biological culture 120. It is used to selectively stimulate the different neurons, dendrites or axons. The Stimulation Unit 110 may control the digital data input signal transfer [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns] to the biological culture, by varying it spatially (addressing different neurons), temporally (stimulating with a variable signal over time), and/or spatio-temporally… The Readout Unit 130 represents the output interface between the biological culture 120 and a digital data output signal 135. It is used to selectively measure the activity of different neurons, dendrites or axons. The Readout Unit 130 may control the biological culture activity conversion to a digital data output signal, possibly multidimensional, by sampling it spatially (capturing the individual activity of different neurons), temporally (capturing a variable signal over time), and/or spatio-temporally…; And in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA) [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns]... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit, using measurement systems such as a multielectrode array (MEA) receiving an electrical signal [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns]. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.)
and using the measured at least one response from the BNN in furtherance of performing the task. (in [0039] A subset of the neurons may be excited with an input signal through a BNN input interface such as the electrical signals of a multielectrode array (MEA)... The electrical activity of the neuron cells may be monitored at several places in the biological material, as the outputs of the BNN unit [measuring, using the MEA, at least one response of the BNN that is responsive to the stimulating with the at least one temporal sequence of stimulation patterns], using measurement systems such as a multielectrode array (MEA) receiving an electrical signal. Alternately, a subset of the neurons may be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system [and using the measured at least one response from the BNN in furtherance of performing the task]. However, there exists numerous other systems enabling or facilitating interfacing a BNN, for instance processes exploiting concepts similar to those in the human body, such as conversion of electrochemical stimuli into mechanical movement and observable in muscle movement or speech.; And in [0093] In a possible embodiment, the BNN functional interface 730 may include various input and/or output signal processing algorithms such as pre-processing or post-processing filters, classifiers, machine learning algorithms based on mathematical or statistical models, and the functional control software 701 may control those algorithms as well as their parameters in accordance with the end user needs. Signal pre-processing may comprise transforming the signals to be learnt by the BNN so that they optimally fit the SU stimulation capability and formats. Signal post-processing may comprise transforming the signals output from the BNN to a more comprehensive format so that they provide the target functionality. In a possible embodiment the BNN may be modeled as a non-linear system and signal pre-processing may comprise applying a non-linear gain to the input signals...)
Additionally, Peter teaches … at least one temporal sequence of two-dimensional stimulation patterns; as depicted in Figure :
And in Pg. 431: Reservoir Computing Artificial neural networks (ANNs) represent a class of computational models that take inspiration from biological neural networks (BNNs). In ANNs, artificial neurons (the computing elements) are typically arranged in layers, i.e. neurons in one layer are connected to neurons in the next layer and information flows in a feed-forward fashion. Artificial recurrent neural networks (RNNs) are a more plausible and realistic model of BNNs, where the connection topology of neurons has recurrencies, i.e. cycles [… at least one temporal sequence of two-dimensional stimulation patterns]. By allowing cycles, the RNN becomes a dynamic system with a self-sustained temporal activation and possesses memory of previous inputs, i.e. the activation state of the network is a function of previous activation states. Such property is known as ”echo state” (Jaeger, 2003). Artificial RNNs are far more powerful than feed-forward ANNs but they are also far more difficult to train, because learning gradients dissipate over time (making it difficult to learn long-range memory dependencies) and network dynamics can lead to bifurcations. It has recently been suggested that both artificial and biological RNNs may be considered as a high-dimensional medium of dynamics with the ability to represent information in a high-dimensional and discriminating space [… at least one temporal sequence of two-dimensional stimulation patterns]…
Peter, Li and Jordan are analogous art because both involve developing systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for using biological neural network as a reservoir of dynamics for modeling and provide control of an embodied agent, as disclosed by Peter with the systems and techniques based on implementing neuronal cell cultures as compute substrates for performing machine learning computations/tasks, as collectively disclosed by Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Peter, Li and Jordan to allow for the inclusion of morphogenetic principles into computer architecture to decrease energy consumption and to allow for growing tailored neural networks with target functionalities, (Peter, Pg. 430: Introduction).
Additionally, Kag teaches the spike pattens as claimed … at least one temporal sequence of two-dimensional stimulation patterns, in [0215] In at least one embodiment, the resulting temporal spike array can be provided to the MEA 105 as input for stimulating the neurons 135. For example, in at least one embodiment, the dimensionality of the latent space may correspond to the dimensionality of the 2D [… at least one temporal sequence of two-dimensional stimulation patterns] grid or 3D space of the MEA 105 in a 1 to 1 fashion. Each spike array may be provided to the MEA 105 as a series temporal stimulation events for the total number of time steps, t [… at least one temporal sequence of two-dimensional stimulation patterns]. In at least one embodiment, the latent layer size and the time steps can be selected to match the dimensionality of the 2D grid [… at least one temporal sequence of two-dimensional stimulation patterns] or 3D space of the MEA 105 in order to stimulate the neurons in a single stimulation event (e.g., 2D grid size=latent layer size*t).)
Jordan, Li, Peter and Kag are analogous art because both involve developing g machine learning systems and algorithms for implementing artificial neural network for computational applications.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of the prior art for developing a system and method for interfacing a computing device with biological neurons as disclosed by Kag with the systems, methods and processes for performing complex cognitive computations/tasks mimicking and extending the biological brain function using artificial neural networks as collectively disclosed by Peter, Li and Jordan.
One of ordinary skill in the arts would have been motivated to combine the disclosed methods disclosed by Kag, Peter, Li and Jordan in order to provide neural systems that can self-organize to behave in a way which maximizes the complexity of information input and adapts the system’s behavior to increase performance in a task over time, (Kag, 0027).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Herzog et al. (NPL: “Transmission of Spatio-Temporal Patterns from Biological to
Artificial Neural Networks by a Multi-Electrode Array”): teaches e recorded signals from a multi-electrode array as a first step. If we can simulate the artificial network fast enough to process the signals from the biological network in real-time, then a bidirectional connection can be established. The artificial network may generate adequate feedback signals to the biological network and become a part of a new recurrent neurointerface.
Kappel (US 20220254177): teaches that a high-dimensional representation of the biology-related language-based search data may be compared to an individual high-dimensional representation of a biology-related image-based data set by calculating a distance between the two high-dimensional representations. The distance (e.g. Euclidean distance or earth mover's distance) between two high-dimensional representations may be calculated with low effort, if the two high-dimensional representations are represented by vectors (e.g. normalized vectors). The calculation of the distance may be repeated for every individual high-dimensional representations of the plurality of biology-related image-based data sets.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to OLUWATOSIN ALABI whose telephone number is (571)272-0516. The examiner can normally be reached Monday-Friday, 8:00am-5:00pm EST..
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael Huntley can be reached at (303) 297-4307. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/OLUWATOSIN ALABI/Primary Examiner, Art Unit 2129