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 .
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.
Claim 16-20 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.
Claim 16 recites “location data that identifies one or more locations at which the particular electronic device was located.” However, “the particular electronic device” lacks antecedent basis. Neither claim 15 nor claim 16 previously introduces or otherwise identifies a particular electronic device. Claim 15 recites a computer-program product embodied in a non-transitory machine-readable storage medium, instructions configured to cause one or more data processors to perform operations, and a training dataset of neural data including measured field potential and measured spiking data. None of these elements is identified as “the particular electronic device.”
The disclosure also does not clarify what constitutes the recited “location data” in the context of the claimed neural training dataset. Claim 16 states that the location data identifies one or more locations at which the particular electronic device was located, but the disclosure does not explain how those locations are determined, whether electronic device is intended to move between locations or instead be used at different fixed locations, or what relationship the identified locations have to the processing of the neural data. Paragraph [0023] refers to a “target location” in connection with reaching a task, paragraph [0129] states that an I/O device may include a location-tracking device such as a GPS receiver, and paragraph [0133] states that components of the computing system 1600 may be positioned in different physical locations. However, none of these passages describes the claimed location data in connection with the neural training dataset or explains how such location data is used in, affects, or is otherwise relevant to the recited processing of the one or more data sets.
Accordingly, it is unclear what device is intended by “the particular electronic device,” what information constitutes the claimed “location data,” how the location data of the electronic device is determined, whether the device is intended to move between locations or be associated with different fixed locations, and how the location information relates to the processing of the neural training data. Clarification is required as to the intended meaning and role of the recited location data.
Claim 17 repeats the same limitation and therefore contains the same indefiniteness. Claims 18-20 depend directly or indirectly from claim 16 and inherit the indefinite limitation.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 2-7, 9-14, and 16-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
101 Subject Matter Eligibility Analysis
Step 1: Claims 2-7, 9-14, and 16-20 are within the four statutory categories (a process, machine, manufacture or composition of matter).
Step 2A Prong One, Step 2A Prong Two, and Step 2B Analysis:
Step 2A Prong One asks if the claim recites a judicial exception (abstract idea, law of nature, or natural phenomenon). If the claim recites a judicial exception, analysis proceeds to Step 2A Prong Two, which asks if the claim recites additional elements that integrate the abstract idea into a practical application. If the claim does not integrate the judicial exception, analysis proceeds to Step 2B, which asks if the claim amounts to significantly more than the judicial exception. If the claim does not amount to significantly more than the judicial exception, the claim is not eligible subject matter under 35 U.S.C. 101.
Claims 2-7 are directed to a method consisting of a series of steps, meaning that it is directed to the statutory category of process. Claims 9-14 and 16-20 are directed to storage mediums and processors which are machines.
Regarding claim 2, claim 2 depends from claim 1 and incorporates the limitations of claim 1. The following claim element is an abstract idea:
augmenting the measured field potential data (This is an abstract idea of a mental process. The limitation involves evaluating and modifying measured field potential data, such as by selecting channels to drop, shifting data in time, or changing the mean value of the data. A person could review the recorded data, decide which channels to omit, determine how data should be shifted in time, or determine how the values should be adjusted to change the mean. These operations involve observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools, and therefore fall within the mental process grouping of abstract ideas. See MPEP 2106.04(a)(2)(III).).
The following claim elements are additional elements inherited from claim 1 which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
Regarding claim 3, claim 3 depends from claim 1 and incorporates the limitations of claim 1. The following claim element is an abstract idea:
determining one or more batches of the measured field potential data (This is an abstract idea of a mental process. The limitation involves reviewing the measured field potential data and organizing or grouping the data into one or more batches. A person could examine the record data, determine how the data should be divided, and group the data into separate sets or batches. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental processes grouping of abstract ideas.).
The following claim elements are additional elements inherited from claim 1 which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
Regarding claim 4, the rejection of claim 3 is incorporated herein. Claim 4 further recites the following abstract ideas:
processing each batch of the measured field potential data to estimate the spiking data (This is an abstract idea of a mental process. The limitation involves applying mathematical modeling and calculations to measured field potential data to derive an estimate of spiking data. A person could review the measured field potential values, evaluate the relationship between the field potential data and spiking data, and perform calculations to determine estimated spiking values. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental process and mathematical concept groupings of abstract ideas.);
comparing the estimated spiking data and the measured spiking data for each batch to determine loss (This is an abstract idea of a mental process and mathematical concept. Ther limitations involve comparing corresponding estimated and measured spiking values and mathematically determining a loss based on the differences between those values. A person could review the estimated and measured data, evaluate the differences between corresponding values, and perform calculations to determine an amount of loss. These operations involve mathematical calculations, as well as observations, evaluations, judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mathematical concepts and mental processes groupings of abstract ideas.); and
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
updating the neural network parameters based on the loss (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 5, claim 5 depends from claim 1 and incorporates the limitations of claim 1. The following claim element is an abstract idea:
transforming the measured field potential data to a standardized dimension (This is an abstract idea of a mental process. The limitation involves reviewing measured field potential data and reorganizing or mapping the data into a standardized dimension. A person could examine values organized by channel or time, determine a common dimension, and arrange or map the values into that dimension using pen and paper or a written table. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements, including elements inherited from claim 1 and additional elements recited in claim 5, are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
using the read-in model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 6, the rejection of claim 5 is incorporated herein. Claim 6 further recites the following abstract ideas:
processing the measured field potential data… to determine an estimate of latent dynamics trajectories (This is an abstract idea of a mental process. The limitation involves analyzing numerical field potential data and applying mathematical relationships to derive estimated latent values over time. A person could review the measured field potential values, evaluate relationships among the values, and perform calculations to determine estimated values representing changes over time. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
Regarding claim 7, the rejection of claim 5 is incorporated herein. Claim 7 further recites the following abstract ideas:
processing the latent dynamics trajectories… to estimate the spiking data as denoised firing rates (This is an abstract idea of a mental process and mathematical concept. The limitation involves analyzing numerical latent dynamics values and applying mathematical relationships to derive estimated firing-rate values. A person could review the latent values, evaluate relationships among the values, and perform calculations to determine estimated firing rates representing the spiking data. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
through a read-out model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).)
Regarding claim 9, claim 9 depends from claim 8 and incorporates the limitations of claim 8. The following claim element is an abstract idea:
augmenting the measured field potential data (This is an abstract idea of a mental process. The limitation involves evaluating and modifying measured field potential data, such as by selecting channels to drop, shifting data in time, or changing the mean value of the data. A person could review the recorded data, decide which channels to omit, determine how data should be shifted in time, or determine how the values should be adjusted to change the mean. These operations involve observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools, and therefore fall within the mental process grouping of abstract ideas. See MPEP 2106.04(a)(2)(III).).
The following claim elements are additional elements inherited from claim 8 which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
one or more processors (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).);
one or more hardware storage devices (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).)
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
Regarding claim 10, claim 10 depends from claim 8 and incorporates the limitations of claim 8. The following claim element is an abstract idea:
determining one or more batches of the measured field potential data (This is an abstract idea of a mental process. The limitation involves reviewing the measured field potential data and organizing or grouping the data into one or more batches. A person could examine the record data, determine how the data should be divided, and group the data into separate sets or batches. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental processes grouping of abstract ideas.).
The following claim elements are additional elements inherited from claim 8 which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
one or more processors (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).);
one or more hardware storage devices (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).)
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
Regarding claim 11, the rejection of claim 10 is incorporated herein. Claim 11 further recites the following abstract ideas:
processing each batch of the measured field potential data to estimate the spiking data (This is an abstract idea of a mental process. The limitation involves applying mathematical modeling and calculations to measured field potential data to derive an estimate of spiking data. A person could review the measured field potential values, evaluate the relationship between the field potential data and spiking data, and perform calculations to determine estimated spiking values. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental process and mathematical concept groupings of abstract ideas.);
comparing the estimated spiking data and the measured spiking data for each batch to determine loss (This is an abstract idea of a mental process and mathematical concept. Ther limitations involve comparing corresponding estimated and measured spiking values and mathematically determining a loss based on the differences between those values. A person could review the estimated and measured data, evaluate the differences between corresponding values, and perform calculations to determine an amount of loss. These operations involve mathematical calculations, as well as observations, evaluations, judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mathematical concepts and mental processes groupings of abstract ideas.); and
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
updating the neural network parameters based on the loss (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 12, claim 12 depends from claim 8 and incorporates the limitations of claim 8. The following claim element is an abstract idea:
transforming the measured field potential data to a standardized dimension (This is an abstract idea of a mental process. The limitation involves reviewing measured field potential data and reorganizing or mapping the data into a standardized dimension. A person could examine values organized by channel or time, determine a common dimension, and arrange or map the values into that dimension using pen and paper or a written table. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements, including elements inherited from claim 8 and additional elements recited in claim 12, are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
one or more processors (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).);
one or more hardware storage devices (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).)
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
using the read-in model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 13, the rejection of claim 12 is incorporated herein. Claim 13 further recites the following abstract ideas:
processing the measured field potential data… to determine an estimate of latent dynamics trajectories (This is an abstract idea of a mental process. The limitation involves analyzing numerical field potential data and applying mathematical relationships to derive estimated latent values over time. A person could review the measured field potential values, evaluate relationships among the values, and perform calculations to determine estimated values representing changes over time. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
Regarding claim 14, the rejection of claim 12 is incorporated herein. Claim 14 further recites the following abstract ideas:
processing the latent dynamics trajectories… to estimate the spiking data as denoised firing rates (This is an abstract idea of a mental process and mathematical concept. The limitation involves analyzing numerical latent dynamics values and applying mathematical relationships to derive estimated firing-rate values. A person could review the latent values, evaluate relationships among the values, and perform calculations to determine estimated firing rates representing the spiking data. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
through a read-out model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).)
Regarding claim 16, claim 16 depends from claim 15 and incorporates the limitations of claim 15. The following claim element is an abstract idea:
augmenting the measured field potential data (This is an abstract idea of a mental process. The limitation involves evaluating and modifying measured field potential data, such as by selecting channels to drop, shifting data in time, or changing the mean value of the data. A person could review the recorded data, decide which channels to omit, determine how data should be shifted in time, or determine how the values should be adjusted to change the mean. These operations involve observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools, and therefore fall within the mental process grouping of abstract ideas. See MPEP 2106.04(a)(2)(III).).
The following claim elements, including elements inherited from claim 15 and additional elements recited in claim 16, are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception
A computer-program product tangibly embodied in a non-transitory machine-readable storage medium (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).);
data processors (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).)
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (The step of “receiving” the training dataset is a generic data receiving operation that has been recognized by the courts as well-understood, routine, and conventional activity. See MPEP 2106.05(d)(II).); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
a dynamics model (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).).
wherein the one or more data sets includes location data that identifies one or more locations at which the particular electronic device was located (This limitation merely specifies the information content of the one or more data sets by identifying location information associated with the electronic device. This limitation does not alter how the judicial exception is performed or impose a meaningful limit on the judicial exception and therefore constitutes insignificant extra-solution activity. See MPEP 2106.05(g).),
electronic device (This is a high-level recitation of generic computer components for performing the abstract idea. See MPEP 2106.05(f).)
Regarding claim 17, the rejection of claim 16 is incorporated herein. Claim 17 further recites the following abstract ideas:
determining one or more batches of the measured field potential data (This is an abstract idea of a mental process. The limitation involves reviewing the measured field potential data and organizing or grouping the data into one or more batches. A person could examine the record data, determine how the data should be divided, and group the data into separate sets or batches. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental processes grouping of abstract ideas.).
processing each batch of the measured field potential data to estimate the spiking data (This is an abstract idea of a mental process. The limitation involves applying mathematical modeling and calculations to measured field potential data to derive an estimate of spiking data. A person could review the measured field potential values, evaluate the relationship between the field potential data and spiking data, and perform calculations to determine estimated spiking values. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mental process and mathematical concept groupings of abstract ideas.);
comparing the estimated spiking data and the measured spiking data for each batch to determine loss (This is an abstract idea of a mental process and mathematical concept. Ther limitations involve comparing corresponding estimated and measured spiking values and mathematically determining a loss based on the differences between those values. A person could review the estimated and measured data, evaluate the differences between corresponding values, and perform calculations to determine an amount of loss. These operations involve mathematical calculations, as well as observations, evaluations, judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools. Therefore, the limitation falls within the mathematical concepts and mental processes groupings of abstract ideas.); and
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
updating the neural network parameters based on the loss (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 18, the rejection of claim 16 is incorporated herein. Claim 18 further recites the following abstract ideas:
transforming the measured field potential data to a standardized dimension (This is an abstract idea of a mental process. The limitation involves reviewing measured field potential data and reorganizing or mapping the data into a standardized dimension. A person could examine values organized by channel or time, determine a common dimension, and arrange or map the values into that dimension using pen and paper or a written table. These operations involve observations, evaluations, and judgements and can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
using the read-in model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).).
Regarding claim 19, the rejection of claim 18 is incorporated herein. Claim 19 further recites the following abstract ideas:
processing the measured field potential data… to determine an estimate of latent dynamics trajectories (This is an abstract idea of a mental process. The limitation involves analyzing numerical field potential data and applying mathematical relationships to derive estimated latent values over time. A person could review the measured field potential values, evaluate relationships among the values, and perform calculations to determine estimated values representing changes over time. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
Regarding claim 20, the rejection of claim 18 is incorporated herein. Claim 20 further recites the following abstract ideas:
processing the latent dynamics trajectories… to estimate the spiking data as denoised firing rates (This is an abstract idea of a mental process and mathematical concept. The limitation involves analyzing numerical latent dynamics values and applying mathematical relationships to derive estimated firing-rate values. A person could review the latent values, evaluate relationships among the values, and perform calculations to determine estimated firing rates representing the spiking data. These operations involve mathematical calculations, as well as observations, evaluations, and judgements that can be practically performed in the human mind with the aid of pen and paper or basic computational tools.).
The following claim elements are additional elements which, taken alone or in combination with the other elements, do not integrate the judicial exception into a practical application nor amount to significantly more than the judicial exception:
through a read-out model (This limitation constitutes mere instructions to apply the abstract idea and insignificant extra-solution activity. See MPEP 2106.05(f) and 2106.05(g).)
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-3, 8-10, and 15 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Ahmadi et al. (NPL: “Inferring entire spiking activity from local field potentials” (Published: 2021)).
Regarding claim 1, Ahmadi discloses:
A computer-implemented method, comprising: receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (Ahmadi, [page 7] “Electrophysiological recordings were obtained from two public neural datasets, herein referred to as dataset I and dataset II. These datasets were recorded from the motor cortex area of three Rhesus macaque monkeys (Macaca mulatta) while performing predefined tasks with 96-channel silicon-based intracortical microelectrode (Utah) array.” [page 8] “LFP was obtained by low-pass filtering the raw neural signal with a 4th-order Butterworth filter at 100 Hz and then downsampling it to 1 kHz.” [page 9] “Both datasets comprise the raw neural signals and pre-processed spikes (detected and sorted spikes).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block).” – teaches neural network datasets obtained from macaque subjects containing measured LFP data and detected and sorted spiking data. Ahmadi further divides each dataset into a training set, validation set, and testing set.); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (Ahmadi, [page 3] “We were also interested in determining whether these findings could hold true when using different algorithms. We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)… The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance. The model’s performance evaluation was assessed using Pearson’s correlation coefficients (CC) metric. CC measures the linear correlation between the actual and inferred spiking activity.” – teaches performing the same inference procedure using deep-learning models, including an LSTM. Ahmadi further teaches training on a designated training set with LFP features as input and ESA, SUA, or MUA spiking features as output, and evaluating the resulting inferred spiking activity. Thus, Ahmadi teaches training a neural network architecture to estimate spiking data from field potential data. Ahmadi’s LSTM corresponds to the claimed dynamics model under the broadest reasonable interpretation, as an LSTM is an RNN-based model and the Specification identifies RNN-based models as example of dynamics models (paragraph [0032]).).
Regarding claim 2, Ahmadi discloses:
The method according claim 1, further comprising: augmenting the measured field potential data (Ahmadi, [page 10] “The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches augmenting the measured field potential data by z-transforming the LFP features to have zero mean and unit variance. Under the broadest reasonable interpretation, changing the mean value of the LFP data constitutes augmentation, consistent with the Specification describing augmentation as including “drop channels, time shift, change mean value, etc.” (paragraph [0034]).).
Regarding claim 3, Ahmadi discloses:
The method according to claim 1, further comprising: determining one or more batches of the measured field potential data (Ahmadi, [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)… The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches dividing each neural dataset into ten non-overlapping contiguous blocks of equal size, including eight blocks used for training, with LFP features serving as model input. Under BRI, these blocks constitute batches of the measured field potential data. Thus, Ahmadi teaches determining one or more batches of the measured field potential data.).
Regarding claim 8, Ahmadi discloses:
A system, comprising: one or more processors; and one or more hardware storage devices having stored thereon computer-executable instructions which are executable by the one or more processors to cause the computing system to perform at least the following (Ahmadi, [page 3] “We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” [page 11] “All the analyses were conducted in Python (v3.6.10).” – Ahmadi performs the disclosed deep learning analysis, including LSTM-based analysis, using Python. Such computer-implemented execution necessarily requires one or more processors executing stored computer instructions and memory or storage containing those instructions. Accordingly, the recited processors, hardware storage devices, and computer-executable instructions area inherent in Ahmadi’s disclosed computer-implemented system.):
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (Ahmadi, [page 7] “Electrophysiological recordings were obtained from two public neural datasets, herein referred to as dataset I and dataset II. These datasets were recorded from the motor cortex area of three Rhesus macaque monkeys (Macaca mulatta) while performing predefined tasks with 96-channel silicon-based intracortical microelectrode (Utah) array.” [page 8] “LFP was obtained by low-pass filtering the raw neural signal with a 4th-order Butterworth filter at 100 Hz and then downsampling it to 1 kHz.” [page 9] “Both datasets comprise the raw neural signals and pre-processed spikes (detected and sorted spikes).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block).” – teaches neural network datasets obtained from macaque subjects containing measured LFP data and detected and sorted spiking data. Ahmadi further divides each dataset into a training set, validation set, and testing set.); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (Ahmadi, [page 3] “We were also interested in determining whether these findings could hold true when using different algorithms. We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)… The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance. The model’s performance evaluation was assessed using Pearson’s correlation coefficients (CC) metric. CC measures the linear correlation between the actual and inferred spiking activity.” – teaches performing the same inference procedure using deep-learning models, including an LSTM. Ahmadi further teaches training on a designated training set with LFP features as input and ESA, SUA, or MUA spiking features as output, and evaluating the resulting inferred spiking activity. Thus, Ahmadi teaches training a neural network architecture to estimate spiking data from field potential data. Ahmadi’s LSTM corresponds to the claimed dynamics model under the broadest reasonable interpretation, as an LSTM is an RNN-based model and the Specification identifies RNN-based models as example of dynamics models (paragraph [0032]).).
Regarding claim 9, Ahmadi teaches all the elements of claim 8, therefore is rejected for the same reasons as those presented for claim 8. The claim recites similar limitations corresponding to claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Regarding claim 10, Ahmadi teaches all the elements of claim 8, therefore it is rejected for the same reasons as those presented for claim 8. The claim recites similar limitations corresponding to claim 3 and is rejected for similar reasons as claim 3 using similar teachings and rationale.
Regarding claim 15, Ahmadi discloses:
A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform operations including: (Ahmadi, [page 3] “We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” [page 11] “All the analyses were conducted in Python (v3.6.10).” – Ahmadi’s Python based implementation of the disclosed deep-learning analysis necessarily requires computer-executable instructions stored on a non-transitory machine-readable storage medium and executed by one or more data processors. Accordingly, the recited computer program product is inherent in Ahmadi’s computer-implemented analysis.):
receiving a training dataset of neural data for at least one subject, the training dataset including measured field potential data and measured spiking data (Ahmadi, [page 7] “Electrophysiological recordings were obtained from two public neural datasets, herein referred to as dataset I and dataset II. These datasets were recorded from the motor cortex area of three Rhesus macaque monkeys (Macaca mulatta) while performing predefined tasks with 96-channel silicon-based intracortical microelectrode (Utah) array.” [page 8] “LFP was obtained by low-pass filtering the raw neural signal with a 4th-order Butterworth filter at 100 Hz and then downsampling it to 1 kHz.” [page 9] “Both datasets comprise the raw neural signals and pre-processed spikes (detected and sorted spikes).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block).” – teaches neural network datasets obtained from macaque subjects containing measured LFP data and detected and sorted spiking data. Ahmadi further divides each dataset into a training set, validation set, and testing set.); and
training a neural network architecture to estimate spiking data from the field potential data, wherein the neural network architecture includes a dynamics model (Ahmadi, [page 3] “We were also interested in determining whether these findings could hold true when using different algorithms. We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)… The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance. The model’s performance evaluation was assessed using Pearson’s correlation coefficients (CC) metric. CC measures the linear correlation between the actual and inferred spiking activity.” – teaches performing the same inference procedure using deep-learning models, including an LSTM. Ahmadi further teaches training on a designated training set with LFP features as input and ESA, SUA, or MUA spiking features as output, and evaluating the resulting inferred spiking activity. Thus, Ahmadi teaches training a neural network architecture to estimate spiking data from field potential data. Ahmadi’s LSTM corresponds to the claimed dynamics model under the broadest reasonable interpretation, as an LSTM is an RNN-based model and the Specification identifies RNN-based models as example of dynamics models (paragraph [0032]).).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 4-7, 11-14, and 16-20 are rejected under the 35 U.S.C. 103 as being unpatentable over Ahmadi, (NPL: “Inferring entire spiking activity from local field potentials” (Published: 2021)) in view of Pandarinath et al. (NPL: “Inferring single-trial neural population dynamics using sequential auto-encoders” (Published: 2017)).
Regarding claim 4, Ahmadi teaches all the elements of claim 3, therefore is rejected for the same reasons as those presented for claim 3. Ahmadi further teaches:
processing each batch of the measured field potential data to estimate the spiking data (Ahmadi, [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)…The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches dividing measured neural data into blocks and processing LFP features as input to produce ESA, SUA, or MUA spiking features as output. The blocks correspond to the recited batches.);
However, Ahmadi does not teach but Ahmadi in view of Pandarinath teaches the following limitations:
comparing the estimated spiking data and the measured spiking data for each batch to determine loss (Ahmadi, [page 10] “CC measures the linear correlation between the actual and inferred spiking activity. In addition, we also evaluated the model performance using another metric called root mean square error (RMSE). It is a measure of the average magnitude of the inference error.” Pandarinath, [page 3] “LFADS is a sequential adaptation of a variational auto-encoder… constructed by maximizing a lower bound on the the likelihood of the observed spiking activity being produced by the generator network…” [page 32] “To generate a mini-batch of gradients, the algorithm then selects a random mini-batch of data from that session and propagates it forward to evaluate the loss.” – Ahmadi compares the inferred spiking activity with the corresponding actual spiking activity and determines an inference error. Pandarinath teaches performing the evaluation on a mini-batch of data to determine a loss. Accordingly, the combined teachings provide comparing the estimated spiking data with the measured spiking data for each batch to determine a loss.); and
updating the neural network parameters based on the loss (Pandarinath, [page 32] “The relevant gradients of the loss are then back-propagated. As a result, all shared parameters (e.g. the generator RNN parameters) are modified with every mini-batch of data…” – Pandarinath teaches backpropagating gradients of the loss and modifying the generator RNN parameters for each mini-batch of data…”).
Accordingly, it would have been obvious to a person of ordinary skill in the art, before the effective filing date of the claimed invention, having a combination of Ahmadi and Pandarinath before them, to incorporate the use of mini-batch loss evaluation and backpropagation for updating neural network parameters, as taught by Pandarinath, into the LFP-based spiking inference method of Ahmadi. One would have been motivated to make such a combination in order to minimize the error between inferred and measured spiking activity during neural network training and update the model parameters based on that error. This would allow the trained neural network to more accurately estimate spiking activity from measured field potential data.
Regarding claim 5, Ahmadi teaches all the elements of claim 1, therefore it is rejected for the same reasons as those presented for claim 1. Ahmadi does not teach but Ahmadi in view of Pandarinath teaches the following limitation:
wherein the neural network architecture includes a read-in model, and the training further includes: transforming the measured field potential data to a standardized dimension using the read-in model (Ahmadi, [page 10] “The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” Pandarinath, [page 31] “Therefore, LFADS allows a different input and output transformation for each recording session to handle the different measurements, but otherwise LFADS models all the data with the same generative model, with shared parameters across all recording sessions…” & “To handle multiple sessions’ data, we modified this practice by introducing a per-session input adaptor matrix,
W
s
s
p
i
k
e
. Then, for the bidirectional encoding RNN for g0 we modified equations 9 and 10, by inputting the linearly transformed spikes… where the dimensions of the matrix is
W
s
s
p
i
k
e
(.) are
F
x
D
s
.” – Ahmadi teaches using measured LFP features as input to the neural network. Pandarinath teaches an input adaptor matrix that linearly transforms session-specific neural input having dimension
D
s
to the common factor dimension
F
, which is independent of the recording session. Under BRI, Pandarinath’s input adaptor matrix corresponds to the recited read-in model, and transforming inputs having different sessions-specific dimensions to the common dimension
F
corresponds to transforming the input data to a standardized dimension. Accordingly, the combined teachings provide transforming measured field potential data to a standardized dimension using a read-in model.).
Regarding claim 6, Ahmadi in view of Pandarinath teaches all the elements of claim 5, therefore is rejected for the same reasons as those presented for claim 5. Ahmadi in view of Pandarinath further teaches:
processing the measured field potential data through the dynamics model to determine an estimate of latent dynamics trajectories (Ahmadi, [page 3] “We next compared the inference of ESA from LFP features to the inference of other types of spiking (SUA and MUA) from LFP features… We were also interested in determining whether these findings could hold true when using different algorithms. We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” Pandarinath, [page 4] “The factors are defined as linear readouts from a dynamical generator (an RNN), via a readout matrix
W
f
a
c
.” [page 5] “From this compressed code, the generator RNN infers the dynamic factors and firing rates of all the recorded neurons across time for the encoded trial… The dynamic factors (third panel) compress the dynamics down to 50 dimensions.” – Ahmadi teaches processing measured LFP features using an LSTM neural network. Pandarinath teaches a dynamic generator RNN whose activity is read out as low-dimensional dynamic factors that are inferred across time. Under BRI, Pandarinath’s dynamical generator RNN corresponds to the recited dynamics model, and the low-dimensional dynamics factors inferred across time corresponds to the recited latent dynamics trajectories. Accordingly, the combined teachings provide processing measured field potential data through a dynamics model to determine an estimate of latent dynamic trajectories.).
Regarding claim 7, Ahmadi in view of Pandarinath teaches all the elements of claim 5, therefore is rejected for the same reasons as those presented for claim 5. Ahmadi in view of Pandarinath further teaches:
wherein the neural network architecture includes a readout model and the training further includes: processing the latent dynamics trajectories through a read-out model to estimate the spiking data as denoised firing rates (Pandarinath, [page 3] “For neuron
i
, the LFADS-inferred firing rate
r
t
,
i
provides a de-noised rate for its observed spiking activity on a trial-by-trial basis. The firing rates are obtained by multiplying a vector of dynamic factors
f
t
by a readout matrix
W
r
a
t
e
and then computing an exponential function of the resulting quantity.” [page 4] “The firing rates are linear readouts from a set of low-dimensional factors
f
t
(50) via a readout matrix
W
r
a
t
e
.”- Pandarinath teaches processing low-dimensional dynamic factors through a readout matrix to determine inferred firing rates. As mapped in claim 6, the dynamics factors correspond to the latent dynamics trajectories. Pandarinath further identifies the inferred firing rates as de-noised rates for the observed spiking activity. Accordingly, the readout matrix corresponds to the recited read-out model, and Pandarinath provides processing the latent dynamics trajectories through the read-out model to estimate the spiking data as denoised firing rates.).
Regarding claim 11, Ahmadi teaches all the elements of claim 10, therefore it is rejected for the same reasons as those presented for claim 10. The claim recites similar limitations corresponding to claim 4 and is rejected for similar reasons as claim 4 using similar teachings and rationale.
Regarding claim 12, Ahmadi teaches all the elements of claim 8, therefore it is rejected for the same reasons as those presented for claim 8. The claim recites similar limitations corresponding to claim 5 and is rejected for similar reasons as claim 5 using similar teachings and rationale.
Regarding claim 13, Ahmadi in view of Pandarinath teaches all the elements of claim 12, therefore it is rejected for the same reasons as those presented for claim 12. The claim recites similar limitations corresponding to claim 6 and is rejected for similar reasons as claim 6 using similar teachings and rationale.
Regarding claim 14, Ahmadi teaches in view of Pandarinath all the elements of claim 12, therefore it is rejected for the same reasons as those presented for claim 12. The claim recites similar limitations corresponding to claim 7 and is rejected for similar reasons as claim 7 using similar teachings and rationale.
Regarding claim 16, Ahmadi teaches all the elements of claim 15, therefore it is rejected for the same reasons as those presented for claim 3. Ahmadi further teaches:
augmenting the measured field potential data (Ahmadi, [page 10] “The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches augmenting the measured field potential data by z-transforming the LFP features to have zero mean and unit variance. Under the broadest reasonable interpretation, changing the mean value of the LFP data constitutes augmentation, consistent with the Specification describing augmentation as including “drop channels, time shift, change mean value, etc.” (paragraph [0034]).).
However, Ahmadi does not teach but Ahmadi in view of Pandarinath teaches the following limitation:
wherein the one or more data sets includes location data that identifies one or more locations at which the particular electronic device was located, and wherein processing the one or more data sets comprises (Pandarinath, [page 11] “We tested this approach using neural activity from monkey M1 and PMd during a center-out instructed-delay reaching task, recorded using linear multi-electrode arrays (monkey P; 24 channel V-probes, Plexon Inc.). We trained one stitched multi-session LFADS model on a combined dataset consisting of 44 recording sessions, which included 38 separate electrode penetration sites and spanned 162 days…” [page 12, Figure 5] “Locations of linear electrode array penetrations in the precentral gyrus from which each dataset was collected.” – Pandarinath teaches neural datasets collected using multi-electrode recording hardware at identified electrode penetration site locations. Under BRI, the neural-recording hardware corresponds to the recited electronic device, and the identified penetration site locations correspond to location data identifying locations at which the device was positioned during collection of the respective datasets. The combined dataset is further processed using a multi-session LFADS model.):
Regarding claim 17, Ahmadi in view of Pandarinath teaches all the elements of claim 16, therefore is rejected for the same reasons as those presented for claim 16. Ahmadi in view of Pandarinath further teaches:
determining one or more batches of the measured field potential data (Ahmadi, [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)… The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches dividing each neural dataset into ten non-overlapping contiguous blocks of equal size, including eight blocks used for training, with LFP features serving as model input. Under BRI, these blocks constitute batches of the measured field potential data. Thus, Ahmadi teaches determining one or more batches of the measured field potential data.).
processing each batch of the measured field potential data to estimate the spiking data (Ahmadi, [page 10] “Each dataset was divided into 10 non-overlapping contiguous blocks of equal size which were then categorised into three sets: training set (8 concatenated blocks), validation set (1 block) and testing set (1 block)…The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” – Ahmadi teaches dividing measured neural data into blocks and processing LFP features as input to produce ESA, SUA, or MUA spiking features as output. The blocks correspond to the recited batches.);
comparing the estimated spiking data and the measured spiking data for each batch to determine loss (Ahmadi, [page 10] “CC measures the linear correlation between the actual and inferred spiking activity. In addition, we also evaluated the model performance using another metric called root mean square error (RMSE). It is a measure of the average magnitude of the inference error.” Pandarinath, [page 3] “LFADS is a sequential adaptation of a variational auto-encoder… constructed by maximizing a lower bound on the the likelihood of the observed spiking activity being produced by the generator network…” [page 32] “To generate a mini-batch of gradients, the algorithm then selects a random mini-batch of data from that session and propagates it forward to evaluate the loss.” – Ahmadi compares the inferred spiking activity with the corresponding actual spiking activity and determines an inference error. Pandarinath teaches performing the evaluation on a mini-batch of data to determine a loss. Accordingly, the combined teachings provide comparing the estimated spiking data with the measured spiking data for each batch to determine a loss.); and
updating the neural network parameters based on the loss (Pandarinath, [page 32] “The relevant gradients of the loss are then back-propagated. As a result, all shared parameters (e.g. the generator RNN parameters) are modified with every mini-batch of data…” – Pandarinath teaches backpropagating gradients of the loss and modifying the generator RNN parameters for each mini-batch of data…”).
Regarding claim 18, Ahmadi in view of Pandarinath teaches all the elements of claim 16, therefore is rejected for the same reasons as those presented for claim 16. Ahmadi in view of Pandarinath further teaches:
wherein the neural network architecture includes a read-in model, and the training further includes: transforming the measured field potential data to a standardized dimension using the read-in model (Ahmadi, [page 10] “The model’s input (LFP features) and output (ESA, SUA, or MUA features) were standardised (i.e. z-transformed) to have zero mean and unit variance.” Pandarinath, [page 31] “Therefore, LFADS allows a different input and output transformation for each recording session to handle the different measurements, but otherwise LFADS models all the data with the same generative model, with shared parameters across all recording sessions…” & “To handle multiple sessions’ data, we modified this practice by introducing a per-session input adaptor matrix,
W
s
s
p
i
k
e
. Then, for the bidirectional encoding RNN for g0 we modified equations 9 and 10, by inputting the linearly transformed spikes… where the dimensions of the matrix is
W
s
s
p
i
k
e
(.) are
F
x
D
s
.” – Ahmadi teaches using measured LFP features as input to the neural network. Pandarinath teaches an input adaptor matrix that linearly transforms session-specific neural input having dimension
D
s
to the common factor dimension
F
, which is independent of the recording session. Under BRI, Pandarinath’s input adaptor matrix corresponds to the recited read-in model, and transforming inputs having different sessions-specific dimensions to the common dimension
F
corresponds to transforming the input data to a standardized dimension. Accordingly, the combined teachings provide transforming measured field potential data to a standardized dimension using a read-in model.).
Regarding claim 19, Ahmadi in view of Pandarinath teaches all the elements of claim 18, therefore is rejected for the same reasons as those presented for claim 18. Ahmadi in view of Pandarinath further teaches:
processing the measured field potential data through the dynamics model to determine an estimate of latent dynamics trajectories (Ahmadi, [page 3] “We next compared the inference of ESA from LFP features to the inference of other types of spiking (SUA and MUA) from LFP features… We were also interested in determining whether these findings could hold true when using different algorithms. We repeated the same procedures but using two different deep learning algorithms: multilayer perceptron (MLP) and long short-term memory (LSTM).” Pandarinath, [page 4] “The factors are defined as linear readouts from a dynamical generator (an RNN), via a readout matrix
W
f
a
c
.” [page 5] “From this compressed code, the generator RNN infers the dynamic factors and firing rates of all the recorded neurons across time for the encoded trial… The dynamic factors (third panel) compress the dynamics down to 50 dimensions.” – Ahmadi teaches processing measured LFP features using an LSTM neural network. Pandarinath teaches a dynamic generator RNN whose activity is read out as low-dimensional dynamic factors that are inferred across time. Under BRI, Pandarinath’s dynamical generator RNN corresponds to the recited dynamics model, and the low-dimensional dynamics factors inferred across time corresponds to the recited latent dynamics trajectories. Accordingly, the combined teachings provide processing measured field potential data through a dynamics model to determine an estimate of latent dynamic trajectories.).
Regarding claim 20, Ahmadi in view of Pandarinath teaches all the elements of claim 18, therefore is rejected for the same reasons as those presented for claim 18. Ahmadi in view of Pandarinath further teaches:
processing the latent dynamics trajectories through a read-out model to estimate the spiking data as denoised firing rates (Pandarinath, [page 3] “For neuron
i
, the LFADS-inferred firing rate
r
t
,
i
provides a de-noised rate for its observed spiking activity on a trial-by-trial basis. The firing rates are obtained by multiplying a vector of dynamic factors
f
t
by a readout matrix
W
r
a
t
e
and then computing an exponential function of the resulting quantity.” [page 4] “The firing rates are linear readouts from a set of low-dimensional factors
f
t
(50) via a readout matrix
W
r
a
t
e
.”- Pandarinath teaches processing low-dimensional dynamic factors through a readout matrix to determine inferred firing rates. As mapped in claim 6, the dynamics factors correspond to the latent dynamics trajectories. Pandarinath further identifies the inferred firing rates as de-noised rates for the observed spiking activity. Accordingly, the readout matrix corresponds to the recited read-out model, and Pandarinath provides processing the latent dynamics trajectories through the read-out model to estimate the spiking data as denoised firing rates.).
Conclusion
The prior art of record and not relied upon is considered pertinent to Applicant’s disclosure:
Hsieh, H. L., Wong, Y. T., Pesaran, B., & Shanechi, M. M. (2019). Multiscale modeling and decoding algorithms for spike-field activity. Journal of neural engineering, 16(1), 016018. – teaches multiscale modeling and decoding of simultaneously recorded spike and local field potential (LFP) activity, including adaptive learning of spike-field model parameters.
Skaar, J. E. W., Stasik, A. J., Hagen, E., Ness, T. V., & Einevoll, G. T. (2020). Estimation of neural network model parameters from local field potentials (LFPs). PLoS computational biology, 16(3), e1007725. – teaches using local field potential data in connection with spiking neural network activity and training a convolutional neural network using LFP-derived input.
Rasch, M. J., Gretton, A., Murayama, Y., Maass, W., & Logothetis, N. K. (2008). Inferring spike trains from local field potentials. Journal of neurophysiology, 99(3), 1461-1476. – teaches using local field potential (LFP) features to infer and predict neuronal spiking activity using machine learning models.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Daravanh Phakousonh whose telephone number is (571)272-6324. The examiner can normally be reached Mon - Thurs 7 AM - 5 PM, Every other Friday 7 AM - 4PM.
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, Li B Zhen can be reached at 571-272-3768. 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.
/Daravanh Phakousonh/Examiner, Art Unit 2121
/Li B. Zhen/Supervisory Patent Examiner, Art Unit 2121
Prosecution Insights