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 .
DETAILED ACTION
This office action is in response to the claims filed on 12/29/2025.
Claims 1-20 are presented for examination.
Response to Argument
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 112:
Applicant’s Argument:
The applicant’s argument regarding the 112 (b) rejection based on the claim amendment filed on 12/29/2025.
Examiner’s Response:
The 112(b) rejection is withdrawn in view of the claim amendment filed on 12/29/2025.
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 101:
Applicant’s Argument:
The applicant’s argument regarding the 101 rejection based on the claim amendment filed on 12/29/2025.
Examiner’s response:
Applicant’s argument regarding the 101 rejection based on the claim amendment filed on 12/29/2025. However, the current claim amendment still not overcome the 101 rejection since the claim does not recite the improvement of the machine learning model nor the improvement of the technology in the field. See the 101 rejection section for more detail.
In reference to applicant’s argument regrading rejections under 35 U.S.C. § 103:
Applicant’s Argument:
The applicant’s argument includes the newly amended limitation filed on 12/29/2025.
Examiner’s Response:
This argument includes the newly amended limitations. It has been fully considered but is moot in view of the new grounds of rejection presented below necessitated by the amendment.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-20 rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. There does not seem to be sufficient description for:
The claim 1 recites “a first qain for a first computation technique and a second qain for a second computation technique, wherein the first computation technique and the second computation technique are for performing a matrix operation based on the matrix”, “the first computation technique based on the first qain satisfyinq a criterion with respect to the second qain; applying, by the processor, the first computation technique for the matrix operation; computing, by the processor, a third qain achieved by the first computation technique; and retraining the machine learning model based on the first qain and the third gain.”. The claim 4 recites “ first computation technique”. The claim 9 recites “second gain includes speedup achieved in applying the first computation technique for the matrix operation.” The claim 10 recites: “ second gain is negative reward…, first computational technique”. However, the specification [Par.0036-0037], “, including the selection of an optimization technique for matrix computations based on attributes of an input. In some embodiments, the instructions stored in the memory are also part of a software program that requires matrix operations as part of its execution. The software program may be, for example, a machine learning program, graph analytics program, image processing program, computer vision program, and/or the like, used in applications such as self-driving cars, search engines, speech recognition, and/or the like. [0037] During execution of the software program, various matrices may be generated, and matrix operations may need to be performed for the various matrices..” The specification only describes matrix computations based on attributes of an input, matrix operations may need to be performed for the various matrices, nowhere in specification describes the first gain for first computation technique and second gain for second computational technique as the claim recites. Therefore, the claims 1 ,4 9, 10 are failing to comply with the written description requirement.
The claims 12, 14, 18, 19 recite substantially similar to claims 1, 4, 9, 10 and are similarly rejected. Dependent claims are rejected for inheriting the new matter of a parent claim.
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: “Is the claim to a process, machine, manufacture or composition of matter?”
Yes, the claim sets of 1-11 and 12-20 are directed to a process and a machine, respectively.
Step 2:
Claim 1 recites:
A computer-implemented method comprising: receiving, by a processor, a matrix as an input; extracting, by the processor, one or more features from the matrix; outputting, by a machine learning model based on the one or more features, a first qain for a first computation technique and a second qain for a second computation technique, wherein the first computation technique and the second computation technique are
Step 2A (1): “Does the claim recite an abstract idea, law of nature, or natural phenomenon?”
The limitations, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitations in the mind but for the recitation of generic computer components. That is, other than reciting “computer,” nothing in the claim element precludes the step from practically being performed in the mind.
-extracting one or more features from the matrix – A user can review a matrix and extract features of a matrix;
wherein the first computation technique and the second computation technique are
selecting, , the first computation technique based on the first qain satisfying a criterion with respect to the second qain; this is a mental process, the human can select the first gain if the that satisfy the condition (observation/Evaluation)
applying, , the
computing, by the processor, a third qain achieved by the first computation technique , this is mathematical concept.
Step 2A (2): “Does the claim recite additional elements that integrate the judicial exception into a practical application?”
This judicial exceptions as recited are not integrated into a practical application. In particular, claim 1 only recites a computer to perform the recited steps. The computer is recited at a high level of granularity (i.e. computing system performing generic computer functions) such that it amounts no more than mere instructions to apply the exception using a generic computer component.
In addition, the “receiving” step is interpreted as insignificant extra-solution activity as mere data gathering.
“outputting, by a machine learning model based on the one or more features, a first qain for a first computation technique and a second qain for a second computation technique, and retraining the machine learning model based on the first qain and the third gain.” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)(3)).
Step 2B: “Does the claim recite additional elements that amount to significantly more than the judicial exception?”
The claim limitations reciting the abstract idea do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using a computer to perform the steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
For example,
“receiving, by a processor, a matrix as an input” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of (data storing or data gathering or data outputting) to a judicial exception do not amount to significantly more than the judicial exception itself .
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory"). (this evidence is applied for data gathering/storing data).
“outputting, by a machine learning model based on the one or more features, a first qain for a first computation technique and a second qain for a second computation technique, and retraining the machine learning model based on the first qain and the third gain.” , “processor” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)(3)).
Therefore, the claim limitation does not include elements that amount to significantly more. Claim 1 is not patent eligible.
Claim 2 recites wherein the matrix is a sparse matrix. The additional elements of dependent claim 2 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 2 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 3 recites wherein the one or more features extracted from the matrix include at least one of a number (M) of rows, a number of columns (N), a number of non-zero (NNZ) values, a number of diagonals (Ndiags), a ratio of diagonals with non- zero values to total diagonals (NTdiagsratio), an average number of non-zero values per row (aver RD), a maximum number of non-zero values per row (maxRD), a minimum number of non-zero values per row (maxRD), a deviation of a number of non-zero values per row (devRD), a ratio of non-zero values in a diagonal data structure (ERDIA), a ratio of non-zero values when entries of the matrix are stored in a dense array in column major order (ERELL), a ratio of non-zero values in a row-packed structure (ERRD), an average different between NNZs of adjacent rows (row-bounce), average difference between NNZs of adjacent columns (col_bounce), density of NNZ in the sparse matrix (d), or average number of non-zero neighbors of an element (mean_neighbor). The additional elements of dependent claim 3 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 3 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 4 recites wherein the first computation technique includes a compute kernel for accelerating the matrix operation. The additional elements of dependent claim 4 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 4 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 5 recites wherein the machine learning model is further trained to select a value of a hyperparameter for performing the matrix operation. The limitation of ‘select a value of a hyperparameter’, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind but for the recitation of generic computer components. The limitation, broadly interpreted, can be interpreted as the mental process of selecting a value.
This judicial exceptions as recited are not integrated into a practical application. In particular, claim 1 only recites a computer to perform the recited steps. The computer is recited at a high level of granularity (i.e. computing system performing generic computer functions) such that it amounts no more than mere instructions to apply the exception using a generic computer component.
The claim limitations reciting the abstract idea do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using a computer to perform the steps amounts to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
Therefore, the claim limitation does not include elements that amount to significantly more. Claim 5 is not patent eligible.
Claim 6 recites wherein the matrix operation includes a sparse matrix by dense matrix multiplication (SpMM). The additional elements of dependent claim 6 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 6 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 7 recites wherein the matrix operation includes a general matrix multiply operation (GeMM). The additional elements of dependent claim 7 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 7 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 8 recites wherein the machine learning model includes a deep reinforcement learning model. The additional elements of dependent claim 8 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 8 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 9 recites wherein the second gain includes speedup achieved in applying the first computation technique for the matrix operation. The additional elements of dependent claim 9 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 9 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 10 recites wherein the second gain is a negative reward in response to a dense matrix operation being faster than applying the first computation technique for the matrix operation. The additional elements of dependent claim 10 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 10 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 11 recites wherein the criterion is maximization of the reward. The additional elements of dependent claim 11 do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. MPEP 2106.05(h)(vi): Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016).
Mere instructions to apply an exception using a generic component cannot provide an inventive concept. Because claim 11 is directed to the abstract idea, it does not add significantly more. The claim is not patent eligible.
Claim 12 recites substantially similar to claim 1 and is similarly rejected.
Claim 13 recites substantially similar to claim 3 and is similarly rejected.
Claim 14 recites substantially similar to claim 4 and is similarly rejected.
Claim 15 recites substantially similar to claim 5 and is similarly rejected.
Claim 16 recites substantially similar to claims 6 and 7 and is similarly rejected.
Claim 17 recites substantially similar to claim 8 and is similarly rejected.
Claim 18 recites substantially similar to claim 9 and is similarly rejected.
Claim 19 recites substantially similar to claim 10 and is similarly rejected.
Claim 20 recites substantially similar to claim 11 and is similarly rejected.
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 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)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-3, 5-13, and 15-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Pool et al. (Patent. No. US11392829– hereinafter,Pool).
Regarding claim 1, Pool teaches a computer-implemented method comprising: receiving, by a processor, a matrix as an input (Pool, [Col.2, lines 26-53], “FIG. 1 illustrates an example environment 100 that can be utilized to implement aspects of the various embodiments. In some embodiments, a user may utilize a client device 102 to provide input, such as data to be processed or instructions to perform an operation, among other such options. The client device 102 can be any appropriate computing device capable of enabling a user to provide instructions or data for processing, such as may include a desktop computer, notebook computer, smart phone, tablet computer, computer workstation, gaming console, and the like…” (Pool, col 7, rows 3-28, receives 2 input matrices);;
extracting, by the processor, one or more features from the matrix (Pool, col 7, rows 3-28, extracts feature based on sparsity constraint; Pool, col 5-6, rows 61-67 and 1-24, sparsity constraint indicates of every set of M elements of a matrix, at most N of them can have non-zero values, where a matrix can be a singular row);
outputting, by a machine learning model based on the one or more features, a first qain for a first computation technique and a second qain for a second computation technique, wherein the first computation technique and the second computation technique are(Pool, col 4, rows 15-45] and ([Col.11, lines 40-5-], “In some embodiments the compression can be leveraged to allow for concurrent processing. For example, two submatrices might each be compressed to 50% their original sizes. This may allow for both matrices to be processed together in order to more fully utilize the available resources. In some embodiments, both submatrices are passed to a function, process, or neural network, and the output of the network is the result for the two matrices.” Examiner’s note, the two matrices output result are corresponding to the first gain and second gain computation technique.
selecting, by the processor, the first computation technique based on the first qain satisfyinq a criterion with respect to the second qain (Col.7, lines 30-62], “FIG. 5 illustrates another view 500 of the example compression-based operation that can be utilized in accordance with various embodiments. In this example, a set of eight value rows 502 from the dense matrix is passed through one or more 8-4 multiplexers 504 to generate a set of four entry rows 508 of non-zero values. The result can then be multiplied by another set 506 of rows from the sparse matrix to produce a result, which can then be passed through a reduction tree to an accumulator 512, which can then assemble the resulting output matrix for the operation. It should be noted that such operations can be utilized with other inputs as well, such as may involve dot products with sparse vectors and other such options.The view in FIG. 5 can correspond to a zoomed view to one of the dot products that can be performed using vectors A and B (aVec and bVec) in order to make up the full matrix operation multiplication illustrated in FIG. 4. The aVec 506 is four elements wide, as is the bVec 508. Since A is sparse, say at a 4/8 sparsity constraint, at most four out of the eight elements will have non-zero values. The full bVec 502 is eight elements wide, but since A is sparse only four elements of aVec may require any of the eight elements in bVec to perform the dot product. Since only four elements of bVec are needed, an 8-to-4 mux can be used to select the four relevant elements from the full bVec 502, using the corresponding index metadata from aVec. The four elements from aVec and the four elements from bVec can then be used to perform the dot product. Such an approach effectively performs an 8×8 dot product for the price of a 4×4 dot product, with a little additional muxing involved. The reduction tree and accumulator 512 can be conventional components used for dot product-related and other such operations. The zero value elements of the matrices can be selected in any appropriate manner as well, such as;
applying, by the processor, the first computation technique for the matrix operation (Pool, [ Col.7, lines 40-63], “he view in FIG. 5 can correspond to a zoomed view to one of the dot products that can be performed using vectors A and B (aVec and bVec) in order to make up the full matrix operation multiplication illustrated in FIG. 4. The aVec 506 is four elements wide, as is the bVec 508. Since A is sparse, say at a 4/8 sparsity constraint, at most four out of the eight elements will have non-zero values. The full bVec 502 is eight elements wide, but since A is sparse only four elements of aVec may require any of the eight elements in bVec to perform the dot product. Since only four elements of bVec are needed, an 8-to-4 mux can be used to select the four relevant elements from the full bVec 502, using the corresponding index metadata from aVec. The four elements from aVec and the four elements from bVec can then be used to perform the dot product. Such an approach effectively performs an 8×8 dot product for the price of a 4×4 dot product, with a little additional muxing involved. The reduction tree and accumulator 512 can be conventional components used for dot product-related and other such operations. The zero value elements of the matrices can be selected in any appropriate manner as well, such as through user input or through a random number generator, among other such options.”
computing, by the processor, a third qain achieved by the first computation technique (Pool, Col.7, lines 28-65 and Col.8, lines 1-20], “FIG. 5 illustrates another view 500 of the example compression-based operation that can be utilized in accordance with various embodiments. In this example, a set of eight value rows 502 from the dense matrix is passed through one or more 8-4 multiplexers 504 to generate a set of four entry rows 508 of non-zero values. The result can then be multiplied by another set 506 of rows from the sparse matrix to produce a result, which can then be passed through a reduction tree to an accumulator 512, which can then assemble the resulting output matrix for the operation. It should be noted that such operations can be utilized with other inputs as well, such as may involve dot products with sparse vectors and other such options…”) Examiner’s note, the result of the non-zero dot product is considered as the third gain.)
and retraining the machine learning model based on the (Pool, [Col. 8, lines 1-20], “In one embodiment, it can be advantageous to satisfy the sparsity constraint across the entire layer's weights or activations in order to maintain an even distribution of work. Such an approach may require intervention by the user, either in pruning the network for deployment or in inducing sparsity in the activations. Experiments with both approaches have shown promising results, simply by fine-tuning pre-trained networks. Favorable results were obtained for CNN pruning, and consistent results were observed for recurrent machine-translation networks when using sparsity-constrained weights during training. In at least some embodiments, improved results can be achieved when training networks from scratch, as the constrained network can converge more quickly and to a higher accuracy than the original network that was not necessarily so constrained. Even if the sparsity constraint is not met throughout an entire layer, allowing dedicated kernels to be used, this instruction can allow for various opportunistic benefits discussed herein. By executing more quickly, warps that happen to meet the constraint through fine-grained, unstructured pruning, or natural sparsity in various activations can be more efficient. Thus, pruning to the sparsity constraint can be considered to be an optional, but useful, step in some embodiments.”). .
Regarding claim 2, Pool teaches the method of claim 1. Pool additionally discloses wherein the matrix is a sparse matrix (Pool, col 4, rows 35-49, sparse matrix).
Regarding claim 3, Pool teaches the method of claim 1. wherein the one or more features extracted from the matrix include at least one of a number (M) of rows, a number of columns (N), a number of non-zero (NNZ) values, a number of diagonals (Ndiags), a ratio of diagonals with non- zero values to total diagonals (NTdiagsratio), an average number of non-zero values per row (aver RD), a maximum number of non-zero values per row (maxRD), a minimum number of non-zero values per row (maxRD), a deviation of a number of non-zero values per row (devRD), a ratio of non-zero values in a diagonal data structure (ERDIA), a ratio of non-zero values when entries of the matrix are stored in a dense array in column major order (ERELL), a ratio of non-zero values in a row-packed structure (ERRD), an average different between NNZs of adjacent rows (row-bounce), average difference between NNZs of adjacent columns (col_bounce), density of NNZ in the sparse matrix (d), or average number of non-zero neighbors of an element (mean_neighbor) (Pool, col 5-6, rows 61-67 and 1-24, sparsity constraint indicates of every set of M elements of a matrix, at most N of them can have non-zero values, where a matrix can be a singular row).
Regarding claim 5, Pool teaches the method of the claim 1, wherein the machine learning model is further trained to select a value of a hyperparameter for performing the matrix operation (Pool, col 18, rows 34-53, hyperparameter optimization).
Regarding the claim 6, Pool teaches the method of the claim 1, wherein the matrix operation includes a sparse matrix by dense matrix multiplication (SpMM) (Pool, col 4, rows 15-34, sparse by dense matrix multiplication, also shown in fig 2a).
Regarding the claim 7, Pool teaches the method of the claim 1, wherein the matrix operation includes a general matrix multiply operation (GeMM) (Pool, col 9, rows 48-67, with regards to fig 5, general matrix to matrix multiplication).
Regarding claim 8, Pool teaches the method of claim 1, wherein the machine learning model includes a deep reinforcement learning model (Pool, col 4, rows 14-34, deep neural network represents a deep reinforcement learning model).
Regarding claim 9, Pool teaches the method of claim 1, wherein the second gain include speedup achieved in applying the fist computation technique for the matrix operation (Pool, col 6, rows 25-45, reduces computational load, therefore speeding up total processes).
Regarding claim 10, Pool teaches the method of claim 1, wherein the second gain is a negative reward in response to a dense matrix operation being faster than applying the first computation technique for the matrix operation (Pool, col 6, rows 25-45, reduces computational load, reduction represents a negative).
Regarding claim 11, Pool teaches the method of claim 1, wherein the criterion is maximization of a reward (Pool, col 19, rows 24-38, optimization loop for models represents a maximization).
Claim 12 recites substantially similar to claim 1 and is similarly rejected.
Claim 13 recites substantially similar to claim 3 and is similarly rejected.
Claim 15 recites substantially similar to claim 5 and is similarly rejected.
Claim 16 recites substantially similar to claims 6 and 7 and is similarly rejected.
Claim 17 recites substantially similar to claim 8 and is similarly rejected.
Claim 18 recites substantially similar to claim 9 and is similarly rejected.
Claim 19 recites substantially similar to claim 10 and is similarly rejected.
Claim 20 recites substantially similar to claim 11 and is similarly rejected.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 4, 14 are rejected under 35 U.S.C. 103 as being unpatentable over Pool United States Patent US 11,392, 829 in view of Gasser United States Patent US 10,719,366.
Regarding claim 4, Pool in view of Gasser discloses the method of claim 1. Gasser additionally discloses wherein the first computation technique includes a compute kernel to be invoked for accelerating the matrix operation. (Gasser, col 4, rows 30-67, selects number of compute kernels based on machine learning model).
Before the time of the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to have modified the user of matrices to include applying it to the circumstance of compute kernel selection based on the teachings of Gasser. The motivation for doing so would have been creating and maintaining dynamic systems (Gasser, col 1 rows 27-41).
Claim 14 recites substantially similar to claim 4 and is similarly rejected.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EM N TRIEU whose telephone number is (571)272-5747. The examiner can normally be reached on Mon-Fri from 9:00-5:00.
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, Omar Fernandez Rivas can be reached on (571) 272-2589. 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.
/E.T./Examiner, Art Unit 2128
/BRIAN M SMITH/Primary Examiner, Art Unit 2122
Prosecution Insights