DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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. The analysis of the claims’ subject matter eligibility will follow the 2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50-57 (January 7, 2019) (“2019 PEG”).
With respect to claim 1.
Claim 1 is 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—claim 1 recites a method, which is a process.
Step 2A, prong one: Does the claim recite an abstract idea, law of nature or natural phenomenon? Yes—the limitations identified below each, under its broadest reasonable interpretation, covers mental processes abstract idea grouping (concepts performed in the human mind (including an observation, evaluation, judgment, opinion)), see MPEP 2106.04(a)(2), subsection III and the 2019 PEG, but for the recitation of generic computer components:
“(a1) adding at least one new fusion to at least one maintained fusion route, wherein each maintained fusion route comprises at least one fusion, and each fusion comprises at least one operation unit (OP) of the plurality of OPs; (a2) calculating a total execution cost of the at least one maintained fusion route after the at least one new fusion is added; (a3) comparing all total execution costs of all maintained fusion routes having a same end OP; and (a4) selecting a maintained fusion route having a lowest total execution cost from all maintained fusion routes having the same end OP, and discarding all other maintained fusion routes having the same end OP.”: (Mental processes- concept of observation and evaluation of adding data, calculating and comparing data and selecting based on a criteria. This is also a mathematical concept of evaluating cost comparing and selecting minimum cost and discarding paths).
Step 2A, prong two: Does the claim recite additional elements that integrate the judicial exception into a practical application? No—the judicial exception is not integrated into a practical application.
“a machining learning architecture, wherein the machining learning architecture comprises a plurality of operation units (OPs)”: mere instructions to “apply it” because it only includes high level of generality description to apply the abstract idea. See MPEP § 2106.05(f). Therefore, the additional element(s) do not integrate the judicial exception into a practical application. See MPEP 2106.05(f).
The generic computer components in these steps are recited at a high-level of generality (i.e., as a generic computer component performing a generic computer function) such that it amounts no more than mere instructions to apply the exception using a generic computer component. Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. The claim is directed to an abstract idea.
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception? No—there are no additional limitations beyond the mental processes identified above. The limitation treated above, are directed to the well-understood, routine, and conventional activity of storing and retrieving information in memory. See MPEP § 2106.05(d)(II); Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015). It also includes limitations that Merely reciting the words “apply it” (or an equivalent) with the judicial exception, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea, as discussed in MPEP § 2106.05(f). The additional element is insignificant application, which is similar to examples of activities that the courts have found to be insignificant extra-solution activity, in accordance with MPEP 2106.05(g), Insignificant Extra-Solution Activity. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept.
Thus, considering the additional elements individually and in combination and the claims as a whole, the additional elements do not provide significantly more than the abstract idea. This claim is not patent eligible.
Claim 2.
Step 1: A method, as above.
Step 2A Prong 1: The claim recites that “determining if there is only one maintained fusion route and the end OP of the only one maintained fusion is the last OP of the plurality of OPs, if yes, selecting the only one maintained fusion route as a target fusion route, otherwise repeating the steps of (a1) to (a4).”: This limitation merely further limits the mental processes- and or mathematical concept of claim 1.
Step 2A Prong 2, Step 2B: This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 3.
Step 1: A method, as above.
Step 2A Prong 1: The claim recites that “generating the at least one maintained fusion route; wherein the at least one maintained fusion route starts from a first OP of the plurality of OPs.”: This limitation merely further limits the mental processes- and or mathematical concept of claim 1.
Step 2A Prong 2, Step 2B: This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 4.
Step 1: A method, as above.
Step 2A Prong 1: The claim recites that “adding a new fusion at the end of the at least one maintained fusion route; wherein the new fusion comprises at least one OP of the plurality of OPs.”: This limitation merely further limits the mental processes- and or mathematical concept of claim 1.
Step 2A Prong 2, Step 2B: This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 5.
Step 1: A method, as above.
Step 2A Prong 1: The claim recites that “adding a first new fusion at the end of a first maintained fusion route, wherein the first new fusion comprises at least one OP of the plurality of OPs; adding a second new fusion at the end of the first maintained fusion route, wherein the second new fusion comprises at least one OP of the plurality of OPs.”: This limitation merely further limits the mental processes- and or mathematical concept of claim 1.
Step 2A Prong 2, Step 2B: This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 6.
Step 1: A method, as above.
Step 2A Prong 2, Step 2B: The claim recites that “wherein an amount of OPs of a new fusion is determined by a basic structure constraint of the machining learning architecture.” involves Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, e.g., a limitation indicating that a particular function such as creating and maintaining electronic records is performed by a computer, as discussed in Alice Corp., 573 U.S. at 225-26, 110 USPQ2d at 1984 (see MPEP § 2106.05(f)). This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 7.
Step 1: A method, as above.
Step 2A Prong 2, Step 2B: The claim recites that “saving information of the at least one maintained fusion route and/or discarded maintained fusion routes to a table for updating a list of the maintained fusion routes.” involves Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, e.g., a limitation indicating that a particular function such as creating and maintaining electronic records is performed by a computer, as discussed in Alice Corp., 573 U.S. at 225-26, 110 USPQ2d at 1984 (see MPEP § 2106.05(f)). This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 8.
Step 1: A method, as above.
Step 2A Prong 2, Step 2B: The claim recites that “wherein the total execution cost comprises at least one of external memory access cost, cycles, latency, and MAC operations, and the total execution cost is calculated according to all fusions of each maintained fusion route” involves Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, e.g., a limitation indicating that a particular function such as creating and maintaining electronic records is performed by a computer, as discussed in Alice Corp., 573 U.S. at 225-26, 110 USPQ2d at 1984 (see MPEP § 2106.05(f)). This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 9.
Step 1: A method, as above.
Step 2A Prong 2, Step 2B: The claim recites that “wherein the machining learning architecture comprises a graph structure.” involves Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, e.g., a limitation indicating that a particular function such as creating and maintaining electronic records is performed by a computer, as discussed in Alice Corp., 573 U.S. at 225-26, 110 USPQ2d at 1984 (see MPEP § 2106.05(f)). This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claim 10.
Step 1: A method, as above.
Step 2A Prong 2, Step 2B: The claim recites that “wherein the machining learning architecture is used for executing a deep learning algorithm.” involves Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, e.g., a limitation indicating that a particular function such as creating and maintaining electronic records is performed by a computer, as discussed in Alice Corp., 573 U.S. at 225-26, 110 USPQ2d at 1984 (see MPEP § 2106.05(f)). This judicial exception is not integrated into a practical application. Mere recitation of generic computer components neither integrates the judicial exception into a practical application nor provides an inventive concept.
Claims 11-20
Step 1: The claims recite a system; therefore, they fall into the statutory category of machines.
Step 2A Prong 1: The claims recite the same mental processes as claims 11-20, respectively.
Step 2A Prong 2: This judicial exception is not integrated into a practical application. Claims 11-20 recite generic computer components, namely “a machining learning architecture, comprising a processor, wherein the machining learning architecture comprises a plurality of operation units (OPs)”. As before, the mere recitation that the method is to be performed on a generic computer amounts to a mere instruction to apply the exception on the computer. See MPEP § 2106.05(f). With that exception, the analysis mirrors that of claims 1-10, respectively.
Step 2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. The analysis, with the one exception noted above, mirrors that of claims 1-10, respectively.
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) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Boehm et al. (“On Optimizing Operator Fusion Plans for Large-Scale Machine Learning in SystemML”, 2 Jan 2018, arXiv:1801.00829 [cs.DB]) in view of Zheng et al. (“FusionStitching: Boosting Memory Intensive Computations for Deep Learning Workloads”, 17 Dec 2021 (this version, v2)], arXiv:2009.10924 [cs.DC]).
Regarding claim 1.
Boehm teaches a method of scheduling a fusion route for a machining learning architecture, wherein the machining learning architecture comprises a plurality of operation units (OPs) (see abstract, “we contribute algorithms for (1) candidate exploration of valid fusion plans, (2) cost-based candidate selection, and (3) code generation of local and distributed operations over dense, sparse, and compressed data. Our experiments in SystemML show end-to end performance improvements with optimized fusion plans of up to 21x compared to hand-written fused operators, with negligible optimization and code generation overhead.”, also see page 2, section 2.1, “As shown in Figure 2, these scripts are parsed into a hierarchy of statement blocks, where blocks are delineated by control flow. Per block, we compile DAGs of high-level opera tors (HOPs).”), the method comprising:
(a1) adding at least one new fusion to at least one maintained fusion route, wherein each maintained fusion route comprises at least one fusion, and each fusion comprises at least one operation unit (OP) of the plurality of OPs (see page 4, section 3, “we enumerate partial fusion plans per operator, which represent local fusion decisions. We describe (1) the representations of partial fusion plans in our central memoization table, and (2) an efficient algorithm for populating this memo table in a single pass over the HOP DAG, including pruning techniques.”, see page 4-5, “OFMC Template Abstraction: As the basis of our candidate exploration algorithm, we define the open-fuse merge-close (OFMC) template abstraction…open(Hop h): Indicates if a new fused operator of this template can be started at HOP h, covering its operation and reading materialized inputs…fuse(Hop h, Hop in): Indicates if an open fused operator (of this template) at the input HOP in can be expanded to its consumer HOP h…merge(Hop h, Hop in): Indicates if an open fused operator (of this template) at the consumer HOP h can be expanded to its input HOP in…close(Hop h): Indicates the close status of the template after the HOP h and its validity.”, also Algorithm 1 OFMC Explore (recursive), also see intro, “Fusion Opportunities: The generation of execution plans has many opportunities, where fused operators—in terms of composite operators for chains of basic operators— can improve performance”, also see page 3, section 2.2, “We generate code via a depth-first template expansion to ensure valid ordering. Such plans consist of CNodes, which are either template or basic operation nodes. Template nodes represent generic fused opera tor skeletons that have a specific data binding and contain a DAG of basic operations that encodes the data flow.”);
(a2) calculating a total execution cost of the at least one maintained fusion route after the at least one new fusion is added (see page 3, section 2.2, “we choose the optimal subset of fusion plans using a time-based cost model.”, also see page 6, section 4.3, “4.3 Cost Model Given a plan assignment q of interesting points (i.e., a boolean vector of materialization decisions), we compute the costs C(Pi|q) of the entire plan partition P with an analytical cost model for DAG-structured fusion plans including sparsity-exploitation and redundant computation…where p is a basic or fused operator defined by q and ˆTw p , ˆTr p, and ˆTc p are estimates for its write, read, and computation times. The read and write time estimates are derived from the size of inputs and outputs, normalized by peak read and write memory bandwidth…Cost Computation via Cost Vectors: The costs of a partition C(Pi|q) are computed recursively with getPlanCost(q,Pi,cp) starting from its roots Ri. Shared reads and CSEs are captured via cost vectors cp per fused operator.”);
(a3) see page 5, section 4, “Given a memo table of partial fusion plans, candidate se lection aims to choose the optimal subset of non-conflicting partial fusion plans. We describe the problem and cost model, as well as introduce our cost-based enumeration algorithm MPSkipEnum…”, section 4.1, “Overall, we aim to find the cost-optimal set of fusion plans with the optimization scope of a single HOP DAG at-a-time and hybrid runtime plans that might include single-node and distributed operations. We define this problem as follows: Definition 1. Candidate Selection Problem: Given an operator DAG G, and a set of partial fusions plans P, find the set of optimal, non-conflicting fusion plans P that applied to G minimizes costs C…”, also see page 7, algorithm 2, “/ plan costing and comparison––C ←getPlanCost(W,Pi,Mi,q,C)”);
and (a4) selecting a maintained fusion route having a lowest total execution cost from all maintained fusion routes having the same end OP, see page 7, “Basic Enumeration: Algorithm 2 shows the basic enumeration approach. We iterate over the linearized search space of all 2|Mi| plans (lines 2-20), create a boolean plan assignment q that represents positive materialization decisions (line 3), cost the plan with getPlanCost (line 17), and maintain the best plan q and its costs C (lines 18-20) as we scan through the search space. Figure 7(a) shows an example search space of |Mi| = 4 interesting points and its 16 plans…”, also see page 5, section 4, and 4.1, also see page 6, “If an interesting point (gi → gj)—i.e., a data dependency— is assigned true, then all partial fusion plans with a reference from gi to gj are considered invalid and ignored for costing. After optimization, we simply remove all invalid plans”).
Boehm do not specifically teach comparing all total execution costs of all maintained fusion routes having a same end OP; and discarding all other maintained fusion routes having the same end OP.
Zheng teaches comparing all total execution costs of all maintained fusion routes having a same end OP; and discarding all other maintained fusion routes having the same end OP (see page 9, “To demonstrate the benefits of FusionStitching over previous work, we compare it with the default TensorFlow implementation and XLA”, also see page 9, section 7.2, “We evaluate the speedup of FusionStitching by comparing inference cost or the training time of one iteration for TF, XLA and FusionStitching with the same batch-size.”, also see page 5, “FusionStitching enumerates grouping strategies, and emulates schedules of every sub-root/root op and launch dimension of the fused kernel. As data reuse requires correct data locality in the reuse scope, schedules do not match data locality requirement are discarded. After estimating the performance of each enumeration with latency-evaluator, FusionStitching selects code generation strategy with the best estimated performance”).
Both Boehm and Zheng pertain to the problem of fusion plans, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Boehm and Zheng to teach the above limitations. The motivation for doing so would be “optimization opportunities for fused operators—in terms of fused chains of basic operators—are ubiquitous. These opportunities include (1) fewer materialized intermediates, (2) fewer scans of input data, and (3) the exploitation of sparsity across chains of operators. Automatic operator fusion eliminates the need for hand-written fused operators and significantly improves performance for complex or previously unseen chains of operations. However, existing fusion heuristics struggle to find good fusion plans for complex DAGs or hybrid plans of local and distributed operations. In this paper, we introduce an optimization framework for systematically reason about fusion plans that considers materialization points in DAGs, sparsity exploitation, different fusion template types, as well as local and distributed operations. In detail, we contribute algorithms for (1) candidate exploration of valid fusion plans, (2) cost-based candidate selection, and (3) code generation of local and distributed operations over dense, sparse, and compressed data. Our experiments in SystemML show end-to end performance improvements with optimized fusion plans of up to 21x compared to hand-written fused operators, with negligible optimization and code generation overhead.” (see Zheng Abstract).
Regarding claim 2.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches further comprising: determining if there is only one maintained fusion route and the end OP of the only one maintained fusion is the last OP of the plurality of OPs, if yes, selecting the only one maintained fusion route as a target fusion route, otherwise repeating the steps of (a1) to (a4) (see page 4, section 3.1, “Our memoization (memo) table consists of a set of groups, where each group represents the output of an operator in the HOP DAG, i.e., a logical subexpression. Each group is identified by the operator ID, has access to its operator meta data, and contains a set of valid partial fusion plans for this operator. A partial fusion plan is called a memo table entry, and can reference other groups to represent fusion decisions…Figure 5 shows the HOP DAG and the related memo table after candidate exploration and pruning (described in Section 3.2). All eight operators are represented by groups in the memo table. The group 11 refers to the final matrix multiplication (binary aggregate ba(+*)), and consists of three memo table entries of type Row.”, also see page 5, “Given a memo table of partial fusion plans, candidate se lection aims to choose the optimal subset of non-conflicting partial fusion plans. We describe the problem and cost model, as well as introduce our cost-based enumeration algorithm MPSkipEnum.”, also algorithm 2, “the best plan q*”, also Algorithm 1 OFMC Explore (recursive), also see page 4, “This memo table then allows for simple costing and fusion by traversing the HOP DAG top down, probing for fusion plans, traversing group references, and determining the input HOPs, from where this process repeats until we reach the leaf HOPs.”).
Zheng also teaches (see page 10, section 7.3 “Table 2 shows the kernel breakdown information, including execution time (T) of memory intensive ops (Mem), compute intensive ops (Math), CPU time (CPU, kernel launch and framework scheduling), CUDA memcpy/memset activities (Cpy) and kernel call times (#)…FusionStitching supports more complex fusion patterns than XLA with effective kernel generation”, also see section 7.4).
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 2.
Regarding claim 3.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches further comprising: generating the at least one maintained fusion route; wherein the at least one maintained fusion route starts from a first OP of the plurality of OPs (see page 4, section 3, “The exploration of candidate fusion plans aims to identify all valid partial fusion plans to provide a common input for different plan selection policies and simplify optimization.”, also section 3.2, “Given a HOP DAG and an empty memo table, we aim to efficiently discover all valid partial fusion plans. We intro duce a bottom-up algorithm that is template-oblivious and populates the memo table in a single pass over the DAG.”, also see algorithm 1, “3: return W 4: // Recursive candidate exploration– 5: for all j in 1 to |gi| do // for all operator inputs 6: ofmcExplore(gj, W) 7: // Open initial operator plans”).
Regarding claim 4.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein adding the at least one new fusion to the at least one maintained fusion route comprises: adding a new fusion at the end of the at least one maintained fusion route; wherein the new fusion comprises at least one OP of the plurality of OPs (see page 5, “merge(Hop h, Hop in): Indicates if an open fused operator (of this template) at the consumer HOP h can be expanded to its input HOP in, i.e., if it can merge with fused operators at the input. An example is the merge of Cell templates into Row templates… we fuse and merge existing partial fusion plans from the operator inputs to the current operator (lines 11-15). This step entails iterating over all distinct template types of all inputs and probing pair-wise fusion conditions.”).
Regarding claim 5.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein adding the at least one new fusion for the at least one maintained fusion route further comprises: adding a first new fusion at the end of a first maintained fusion route, wherein the first new fusion comprises at least one OP of the plurality of OPs; adding a second new fusion at the end of the first maintained fusion route, wherein the second new fusion comprises at least one OP of the plurality of OPs (see page 5 and algorithm 1 createPlans(gi,gj,t), “In case of a valid opening condition, we add this memo entry and enumerate merge plans with createPlans. This merging is important to cover scenarios such as X (y z), where the matrix-vector multiplication with X opens a Row template, which can also merge Cell templates over y z.”, also see figure 5 on page 4, example, “The group 11 refers to the final matrix multiplication (binary aggregate ba(+*)), and consists of three memo table entries of type Row. These entries encode fusion alternatives: (1) fuse right R(-1,9), (2) fuse left R(10,-1), and (3) fuse both R(10,9).”).
Zheng also teaches (see page 10, section 7.3 “Table 2 shows the kernel breakdown information, including execution time (T) of memory intensive ops (Mem), compute intensive ops (Math), CPU time (CPU, kernel launch and framework scheduling), CUDA memcpy/memset activities (Cpy) and kernel call times (#)…FusionStitching supports more complex fusion patterns than XLA with effective kernel generation”, also see section 7.4).
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 5.
Regarding claim 6.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein an amount of OPs of a new fusion is determined by a basic structure constraint of the machining learning architecture (see page 4, “OFMC Template Abstraction: As the basis of our candidate exploration algorithm, we define the open-fuse merge-close (OFMC) template abstraction: • open(Hop h): Indicates if a new fused operator of this template can be started at HOP h, covering its operation and reading materialized inputs. For example, the condition of an Outer template is an outer-product-like matrix multiplication with size constraints.”, also see page 5, “where Z is a set of constraints such as memory budgets”).
Zheng also teaches (see page 10, section 7.3 “Table 2 shows the kernel breakdown information, including execution time (T) of memory intensive ops (Mem), compute intensive ops (Math), CPU time (CPU, kernel launch and framework scheduling), CUDA memcpy/memset activities (Cpy) and kernel call times (#)…FusionStitching supports more complex fusion patterns than XLA with effective kernel generation”, also see section 7.4).
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 6.
Regarding claim 7.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches further comprising: saving information of the at least one maintained fusion route and/or discarded maintained fusion routes to a table for updating a list of the maintained fusion routes (see page 5, “ W[gi] ← W[gi]∪createPlans(gi,gj,t)… 21: // Prune redundant plans and memoize–22: pruneRedundant(W, gi)”, i.e. maintaining candidate list and pruning as candidate proceeds).
Regarding claim 8.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein the total execution cost comprises at least one of external memory access cost, cycles, latency, and MAC operations, and the total execution cost is calculated according to all fusions of each maintained fusion route (see page 6, section 4.3, “we compute the costs C(Pi|q) of the entire plan partition P with an analytical cost model for DAG-structured fusion plans including sparsity-exploitation and redundant computation as follows:
PNG
media_image1.png
52
418
media_image1.png
Greyscale
where p is a basic or fused operator defined by q and ˆTw p , ˆTr p, and ˆTc p are estimates for its write, read, and computation times. The read and write time estimates are derived from the size of inputs and outputs, normalized by peak read and write memory bandwidth. For example, reading a 100M×10 dense input matrix at 32GB/s peak read bandwidth, gives us a time estimate of ˆTr p = 1G · 8B/32GB/s = 0.25s. Similarly, the compute time is derived from the number of re quired floating point operations and peak compute band width”).
Regarding claim 9.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein the machining learning architecture comprises a graph structure (see page 2, section 2.1, “As shown in Figure 2, these scripts are parsed into a hierarchy of statement blocks, where blocks are delineated by control flow. Per block, we compile DAGs of high-level opera tors (HOPs).”, also section 4.1, “Overall, we aim to find the cost-optimal set of fusion plans with the optimization scope of a single HOP DAG at-a-time and hybrid runtime plans that might include single-node and distributed operations. We define this problem as follows: Definition 1. Candidate Selection Problem: Given an operator DAG G, and a set of partial fusions plans P, find the set of optimal, non-conflicting fusion plans P that applied to G minimizes costs C…”,).
Regarding claim 10.
Boehm and Zheng teaches the method of claim 1,
Boehm further teaches wherein the machining learning architecture is used for executing a deep learning algorithm (see introduction, “Large-scale ML applications range from data intensive, traditional classification, regression, and clustering use cases, to compute-intensive matrix factorization and deep learning architectures.”).
Claims 11-20 recites a system to perform the method recited in claims 1-10. Therefore the rejection of claims 1-10 above applies equally here.
Related prior arts:
SUGIYAMA et al. (US 20210056488 A1) teaches machine learning apparatus totals up all of the unit work operations included in the plurality of different work processes and judges if the plurality of unit work operations of the same type are similar to each other. The machine learning apparatus defines a similar first unit work operation and second unit work operation as a set of similar work operations, uses a common machine learning algorithm so as to generate a similar work learning model, and performs learning relating to a first work process including the first unit work operation and a second work process including the second unit work operation based on the similar work learning model.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IMAD M KASSIM whose telephone number is (571)272-2958. The examiner can normally be reached 10:30AM-5:30PM, M-F (E.S.T.).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Michael J. Huntley can be reached at (303) 297 - 4307. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/IMAD KASSIM/Primary Examiner, Art Unit 2129
Prosecution Insights