DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . This action is responsive to the Application filed on 04/11/2024. Claims 1-22 are pending in the case. Claims 1, 11, and 17 are independent claims.
Claim Rejections - 35 U.S.C. § 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 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 of this title, 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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant are advised of the obligation under 37 C.F.R. § 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. § 102(b)(2)(C) for any potential 35 U.S.C. § 102(a)(2) prior art against the later invention.
Claims 1, 3, 5, and 7-10 are rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan et al. (Sullivan, Jonathan, Arman Mirhashemi, and Jaeho Lee. "Deep learning based analysis of microstructured materials for thermal radiation control." Scientific reports 12, no. 1 (2022): 9785, hereinafter Sullivan) in view of Lenaerts et al. (Lenaerts, Joeri, Hannah Pinson, and Vincent Ginis. "Artificial neural networks for inverse design of resonant nanophotonic components with oscillatory loss landscapes." Nanophotonics 10, no. 1 (2020): 385-392, hereinafter Lenaerts).
As to independent claim 1, Sullivan teaches:
An apparatus comprising:
a processing device configured to (Abstract. Figure 1.):
receive device design parameters corresponding to a device (Figure 1, Xspan, Zspan, and tsub);
receive operation parameters corresponding to the device (Figure 1, injection wavelength λ);
receive device throughput characteristics from a physics solver (Abstract. Page 2, Results. Page 3, "a FDTD simulation, each frequency/wavelength point we solve at has a corresponding set of optical properties (ε, R, T) so to emulate that behavior we utilize a single wavelength point as a network input");
generate device throughput predictions utilizing the device design parameters, the operation parameters, and an artificial neural network (Figure 1, neural network maps geometry, wavelength, and material data to predicted transmissivity and reflectivity);….
Sullivan does not appear to expressly teach generate a loss gradient utilizing the device throughput characteristics and the device throughput predictions; and train the artificial neural network, utilizing the loss gradient, to generate different device throughput predictions.
Lenaerts teaches generate a loss gradient utilizing the device throughput characteristics and the device throughput predictions (Figure 1. Page 387, "the mean squared error (MSE) between this output and the actual transmission spectrum is the loss function"); and train the artificial neural network, utilizing the loss gradient, to generate different device throughput predictions (Figure 1, Page 387, "the gradient of this loss function is propagated back through the network to update the values of the weights").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the machine learning techniques of Lenaerts to avoid getting stuck in local minima far away from the global optimum (see Lenaerts at page 386).
As to dependent claim 3, Sullivan further teaches wherein the device design parameters include a designed wavelength (Figure 1, "injection wavelength (λ)").
As to dependent claim 5, Sullivan further teaches the operation parameters include an operating wavelength (Page 9, "The injection wavelength spans a linearly spaced vector of 100 wavelength points that begins with λmin and ends with λmax"
As to dependent claim 7, Sullivan further teaches generate the loss gradient by comparing the device throughput characteristics and the device throughput predictions (Figure 2(a) compares neural network predicted optical characteristics against FDTD derived optical characteristics and the error values are calculated).
As to dependent claim 8, Sullivan further teaches the physics solver comprises machine-readable instructions executable generate the device throughput characteristics (Page 10, "All datasets used by the neural network are derived from FDTD simulation inputs and outputs directly").
As to dependent claim 9, Sullivan further teaches the physics solver is configured to: receive the design parameters (Figure 1, the FDTD model and the DNN operate on corresponding geometry, material, and wavelength parameters); and receive the operation parameters, wherein the design parameters and the operation parameters are concurrently received by the physics solver and the processing device (Page 3, "The solution to Maxwell’s equations is not sequentially dependent, meaning that we can separate a large, simulated wavelength spectrum into smaller groupings of inputs for the neural network").
As to dependent claim 10, Sullivan further teaches the physics solver is further configured to: generate the device throughput characteristics using the design parameters and the operation parameters (Figure 1, FDTD is supplied geometry, material, and wavelength information and computes corresponding transmissivity and reflectivity); and provide the device throughput characteristics to the processing device (Page 2, "Our model is trained, validated, and tested on a dataset constructed of data compiled from 35,500 different simulations from Lumerical’s commercially available 2D/3D FDTD solver. The simulation framework provides exact solutions for Maxwell’s equations across a finite element mesh and we can extract the dispersion and absorption from the results").
Claims 2 and 6 are rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan in view of Lenaerts and Mehrabian et al. (U.S. Pat. App. Pub. No. 2020/0019851, hereinafter Mehrabian).
As to dependent claim 2, the rejection of claim 1 is incorporated.
Sullivan does not appear to expressly teach the device design parameters include a designed signal intensity.
Mehrabian teaches the device design parameters include a designed signal intensity (Paragraph 37, "the maximum allowed optical power flowing in each physical channel of a photonic accelerator (PA) (500) is bound by the optical power that would produce non-linearities in the silicon waveguides (e.g., 480) and the minimum power that a photo-detector (not shown) can distinguish from noise is when the signal-to-noise ratio of the PA (500) is unity (SNR=1)").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
As to dependent claim 6, the rejection of claim 1 is incorporated.
Sullivan does not appear to expressly teach the device design parameters and the operation parameters correspond to the device comprising a photonic accelerator.
Mehrabian teaches the device design parameters and the operation parameters correspond to the device comprising a photonic accelerator (Figure 5. Paragraph 9, "FIG. 5 illustrates a photonic accelerator").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
Claim 4 is rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan in view of Lenaerts and Abdelli et al. (Abdelli, Khouloud, Danish Rafique, Helmut Grießer, and Stephan Pachnicke. "Lifetime prediction of 1550 nm DFB laser using machine learning techniques." In Optical Fiber Communication Conference, pp. Th2A-3. Optica Publishing Group, 2020, hereinafter Abdelli).
As to dependent claim 4, the rejection of claim 1 is incorporated.
Sullivan does not appear to expressly teach the operation parameters include an operating signal intensity.
Abdelli teaches the operation parameters include an operating signal intensity (Figure 4, "operating conditions like the accelerated aging tests (same optical power Pop = 10 mW, different temperatures)").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
Claims 11-13, 17-20, and 22 are rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan in view of Mehrabian and Abdelli.
As to independent claim 11, Sullivan teaches
A method, comprising:
generating device throughput characteristics for a device… using a first artificial neural network (ANN) (Page 3, "The output of the neural network is the reflectivity and transmissivity that correspond to the wavelength input and material/geometric properties. This design emulates the output of the power monitors used in the FDTD simulations");….
Sullivan does not appear to expressly teach a device comprising a photonic accelerator.
Mehrabian teaches a device comprising a photonic accelerator (Figure 5, photonic accelerator (PA) 500).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
Sullivan does not appear to expressly teach generating, using a mean time to failure (MTTF) heuristic, a first MTTF prediction using the device throughput characteristics, device design parameters, and operation parameters; generating a second MTTF prediction using the device throughput characteristics, and a second ANN; generating a loss feedback using the first MTTF prediction and the second MTTF prediction; and training the second ANN using the loss feedback.
Abdelli teaches generating, using a mean time to failure (MTTF) heuristic, a first MTTF prediction using the device throughput characteristics, device design parameters, and operation parameters (Figure 2, dataset generation process. Page 2, Equation 1, MTTF is analytically estimated based on optical and electrical characteristics and operating stress parameters, including optical power, current, wavelength, and temperature); generating a second MTTF prediction using the device throughput characteristics, and a second ANN (Page 2, section 2.2 artificial neural network model, ANN receives device operating characteristics and is trained to predict MTTF); generating a loss feedback using the first MTTF prediction and the second MTTF prediction (Page 2, "A mean square error cross-entropy function was used as the loss function to update the weights of the model based on the error between the predicted and the desired output"); and training the second ANN using the loss feedback (Page 2, "the ANN model was trained using the backpropagation algorithm").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
As to dependent claim 12, Sullivan further teaches the first ANN is a deep neural network (Title. Page 1, "Deep-Neural Networks (DNNs)").
As to dependent claim 13, Sullivan further teaches receiving, at the deep neural network, the device design parameters and the operation parameters (Figure 1 supplies numerical geometry, wavelength, and material data as DNN inputs).
Sullivan does not appear to expressly teach which are in a digital format.
Mehrabian teaches which are in a digital format (Figure 5, values are stored in memory buffers and then are passed through digital-to-analog converters before being supplied to the photonic hardware).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
As to independent claim 17, Sullivan teaches
… generate device throughput characteristics using a first artificial neural network (ANN), wherein the device throughput characteristics correspond to a device… (Page 3, "The output of the neural network is the reflectivity and transmissivity that correspond to the wavelength input and material/geometric properties. This design emulates the output of the power monitors used in the FDTD simulations");…
Sullivan does not appear to expressly teach a non-transitory machine-readable medium having computer-readable instructions, which when executed by a computer, cause the computer to; device comprising a photonic accelerator; store the device throughput characteristics in memory; and operation parameters, accessed from the memory and device design parameters accessed from a different memory.
Mehrabian teaches a non-transitory machine-readable medium having computer-readable instructions, which when executed by a computer, cause the computer to (Paragraph 54); device comprising a photonic accelerator (Figure 5, photonic accelerator (PA) 500); store the device throughput characteristics in memory (Claim 6); and operation parameters, accessed from the memory and device design parameters accessed from a different memory (Figure 5, off-chip digital memory 580, separate buffers, and an analog memory 550, kernel and input data follow distinct storage and processing paths).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
Sullivan does not appear to expressly teach generate, using a mean time to failure (MTTF) heuristic, a first MTTF prediction using the device throughput characteristics, and operation parameters… wherein the first MTTF prediction, the device design parameters, and the operation parameters correspond to the device comprising the photonic accelerator; and store the first MTTF prediction in the memory to make the first MTTF prediction available to train a second ANN.
Abdelli teaches generate, using a mean time to failure (MTTF) heuristic, a first MTTF prediction using the device throughput characteristics, and operation parameters… wherein the first MTTF prediction, the device design parameters, and the operation parameters correspond to the device comprising the photonic accelerator (Figure 2 and equation 1 analytical MTTF is determined from laser characteristics and operating parameters); and store the first MTTF prediction in the memory to make the first MTTF prediction available to train a second ANN (Figure 2, laser lifetime prediction dataset).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
As to dependent claim 18, Abdelli further teaches the instructions are further executable to generate a second MTTF prediction using the device throughput characteristics accessed from the memory and the second ANN (Page 3, "a ML approach based on ANN for laser MTTF prediction. Synthetic data including different electro-optical laser characteristics is used to evaluate the prediction accuracy of the proposed model").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
As to dependent claim 19, Abdelli further teaches the instructions are further executable to generate a loss feedback using the first MTTF prediction accessed from memory and the second MTTF prediction (Page 2, "the loss function to update the weights of the model based on the error between the predicted and the desired output").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
As to dependent claim 20, Abdelli further teaches the instructions are further executable to train the second ANN using the loss feedback (Page 2, "the ANN model was trained using the backpropagation algorithm").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
As to dependent claim 22, Sullivan further teaches the instructions are further executable to store a plurality of device throughput characteristics including the device throughput characteristics (Page 2, "finite-difference time-domain (FDTD) simulation outputs by predicting spectral properties for a combination of the plane-wave source wavelength, geometric properties of the texture, and material").
Sullivan does not appear to expressly teach a plurality of operation parameters including the operation parameters, wherein the plurality of device throughput characteristics and the plurality of operation parameters correspond to a plurality of different devices.
Abdelli teaches a plurality of operation parameters including the operation parameters, wherein the plurality of device throughput characteristics and the plurality of operation parameters correspond to a plurality of different devices (Section 2.1, Data Generation).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the lifetime prediction machine learning techniques of Abdelli to have a prediction method that is more accurate and better at unseen operating conditions (see Abdelli at section 1. Introduction).
Claims 14-16 are rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan in view of Mehrabian, Abdelli, and Banerjee et al. (U.S. Pat. App. Pub. No. 2022/0222522, hereinafter Banerjee).
As to dependent claim 14, the rejection of claim 13 is incorporated.
Sullivan does not appear to expressly teach the second ANN is a spiking neural network.
Banerjee teaches the second ANN is a spiking neural network (Paragraphs 3-5. Figure 2, SNN 212).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the spiking neural network techniques of Banerjee for energy and data efficient edge computing (see Banerjee at paragraph 3).
As to dependent claim 15, Sullivan does not appear to expressly teach receiving, at the spiking neural network, the device throughput characteristics in an analog format.
Banerjee teaches receiving, at the spiking neural network, the device throughput characteristics in an analog format (Abstract, " the real world measurements/input signals are in analog (continuous or discrete) signal format." Paragraph 31, "Encoder 206 configured for generating a plurality of initial encoded spike trains").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the spiking neural network techniques of Banerjee for energy and data efficient edge computing (see Banerjee at paragraph 3).
As to dependent claim 16, Sullivan does not appear to expressly teach converting the device throughput characteristics to the analog format utilizing a software spike generator.
Banerjee teaches converting the device throughput characteristics to the analog format utilizing a software spike generator (Figure 2, Encoder 206. Paragraph 31, "Encoder 206 configured for generating a plurality of initial encoded spike trains").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the spiking neural network techniques of Banerjee for energy and data efficient edge computing (see Banerjee at paragraph 3).
Claim 21 is rejected under 35 U.S.C. § 103 as being unpatentable over Sullivan in view of Mehrabian, Abdelli, and Xu et al. (Xu, Qing, Zhenghua Chen, Keyu Wu, Chao Wang, Min Wu, and Xiaoli Li. "KDnet-RUL: A knowledge distillation framework to compress deep neural networks for machine remaining useful life prediction." IEEE Transactions on Industrial Electronics 69, no. 2 (2021): 2022-2032, hereinafter Xu).
As to dependent claim 21, the rejection of claim 20 is incorporated.
Sullivan does not appear to expressly teach different device comprising a different photonic accelerator.
Mehrabian teaches different device comprising a different photonic accelerator (Figure 5, photonic accelerator PA 500).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the optical neural network techniques of Mehrabian to reduce power consumption (see Mehrabian at paragraph 24).
Sullivan does not appear to expressly teach the trained second ANN is executable by a different computer to generate a different MTTF prediction for a different device.
Xu teaches the trained second ANN is executable by a different computer to generate a different MTTF prediction for a different device (Abstract, "the RUL prediction algorithms are required to be deployed on edge devices." Page 10, "transfer the knowledge learned from dataset A to the new machine B without collecting labeled data from machine B, which is also known as domain adaptation").
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having the photonic circuits of Sullivan to include the remaining useful life prediction techniques of Xu to improve the reliability of industrial systems and reduce maintenance cost (see Xu at abstract).
Citation of Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Lu et al. (U.S. Pat. No. 11,550,971) teaches configuring a simulated environment to be representative of a physical device based, at least in part, on an initial description of the physical device that described structural parameters of the physical device. The operations further comprise performing a physics simulation with an artificial intelligence (“AI”) accelerator. The AI accelerator includes a matrix multiply unit for computing convolution operations via a plurality of multiply-accumulate units. The operations further comprise computing a field response in response of the physical device in response to an excitation source within the simulated environment when performing the physics simulation. The field response is computed, at least in part, with the convolution operations to perform spatial differencing.
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicant's disclosure. Applicant is required under 37 C.F.R. § 1.111(c) to consider these references fully when responding to this action.
It is noted that any citation to specific pages, columns, lines, or figures in the prior art references and any interpretation of the references should not be considered to be limiting in any way. A reference is relevant for all it contains and may be relied upon for all that it would have reasonably suggested to one having ordinary skill in the art. In re Heck, 699 F.2d 1331, 1332-33, 216 U.S.P.Q. 1038, 1039 (Fed. Cir. 1983) (quoting In re Lemelson, 397 F.2d 1006, 1009, 158 U.S.P.Q. 275, 277 (C.C.P.A. 1968)).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Casey R. Garner whose telephone number is 571-272-2467. The examiner can normally be reached Monday to Friday, 8am to 5pm, Eastern Time.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Alexey Shmatov can be reached on 571-270-3428. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center and the Private Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from Patent Center or Private PAIR. Status information for unpublished applications is available through Patent Center and Private PAIR to authorized users only. Should you have questions about access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
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) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/Casey R. Garner/Primary Examiner, Art Unit 2123
Prosecution Insights