DETAILED ACTION
This action is responsive to the application filed on 09/26/2024. Claims 1-12 are pending and have been examined. This action is Non-final.
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 .
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C.
120, 121, 365(c), or 386(c) is acknowledged.
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 1-12 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1,
Step 1: The claim is directed to a non-transitory computer-readable recording medium, which falls under the category of manufacture. The claim satisfies step 1.
Step 2A Prong 1:
“classifying the plurality of tasks into a plurality of groups based on similarities of one or more pipelines selected based on evaluation values, among the plurality of pipelines, and similarities of evaluation values of the one or more pipelines” -- The limitation of classifying tasks into groups, under its broadest reasonable interpretation, covers a person observing the pipelines and evaluation values obtained for each task, selecting the best-evaluated pipelines, comparing the selected pipelines and their evaluation values across tasks, and judging which tasks are similar enough to be placed in the same group (e.g., placing tasks having the same top N pipelines in the same group, as described in the specification). This limitation is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
“generating a task group by selecting one task from each of the plurality of groups based on an execution time of the AutoML processing of each of the plurality of tasks” -- The limitation of generating a task group, under its broadest reasonable interpretation, covers a person observing the execution times of the tasks within each group and judging which one task to select from each group (e.g., the task having the shortest execution time), and listing the selected tasks as a task group. This limitation is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“A non-transitory computer-readable recording medium storing a task control program for causing a computer to execute a process comprising” -- The limitation, under its broadest reasonable interpretation, recites the use of a generic computer (a computer, and a program stored on a generic recording medium) to perform the abstract idea. The computer is recited at a high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer. Accordingly, this additional element does not integrate the judicial exception into a practical application, and does not amount to significantly more than the judicial exception (see MPEP 2106.05(f)).
“executing automated machine learning (AutoML) processing on each of a plurality of tasks to acquire a plurality of pipelines for each of the plurality of tasks” -- The limitation recites executing AutoML processing to acquire pipelines at a high level of generality, without any detail as to how the AutoML processing is performed or how the pipelines are generated, and recites only the outcome of the AutoML processing (acquiring a plurality of pipelines). The AutoML processing is thus used as a tool to obtain the pipelines and evaluation values on which the abstract idea operates, and amounts to no more than mere instructions to apply the judicial exception using a generic computer and generic machine learning. Accordingly, this additional element does not integrate the judicial exception into a practical application, and does not amount to significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 1 is non-patent eligible. Claims 5 and 9 are analogous to claim 1, aside from claim type and minute differences, and thus the same rejection applies as above.
Regarding claim 2,
Step 1: The claim is directed to a non-transitory computer-readable recording medium, which falls under the category of manufacture. The claim satisfies step 1.
Step 2A Prong 1:
“The non-transitory computer-readable recording medium according to claim 1, wherein the one task is a task that has a smallest total value of the execution times in a group to which the one task belongs, among the plurality of groups” -- The limitation, under its broadest reasonable interpretation, covers a person totaling the execution times of each task within a group, comparing the totals, and judging which task has the smallest total value. This limitation is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
There are no elements to be evaluated under Step 2A Prong 2 and Step 2B.
Therefore, claim 2 is non-patent eligible. Claims 6 and 10 are analogous to claim 2, aside from claim type and minute differences, and thus the same rejection applies as above.
Regarding claim 3,
Step 1: The claim is directed to a non-transitory computer-readable recording medium, which falls under the category of manufacture. The claim satisfies step 1.
Step 2A Prong 1:
“The non-transitory computer-readable recording medium according to claim 1, wherein … further classifying the plurality of groups into a plurality of first sub-groups based on a deviation between the evaluation values in the one or more pipelines” -- The limitation, under its broadest reasonable interpretation, covers a person observing the evaluation values of the pipelines of the tasks in each group, comparing the deviation between the evaluation values (e.g., noting whether each evaluation value is better or worse than a threshold value), and judging which tasks are to be placed in the same sub-group. This limitation is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“the computer is caused to execute a process of” -- The limitation, under its broadest reasonable interpretation, recites the use of a generic computer (a computer, and a program stored on a generic recording medium) to perform the abstract idea. The computer is recited at a high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer. Accordingly, this additional element does not integrate the judicial exception into a practical application, and does not amount to significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 3 is non-patent eligible. Claims 7 and 11 are analogous to claim 3, aside from claim type and minute differences, and thus the same rejection applies as above.
Regarding claim 4,
Step 1: The claim is directed to a non-transitory computer-readable recording medium, which falls under the category of manufacture. The claim satisfies step 1.
Step 2A Prong 1:
“The non-transitory computer-readable recording medium according to claim 3, wherein…further classifying the plurality of first sub-groups into a plurality of second sub-groups based on degrees of similarity of hyper parameters of a machine learning model” -- The limitation, under its broadest reasonable interpretation, covers a person observing the hyper parameter values of a machine learning model for the tasks in each first sub-group, comparing the hyper parameter values (e.g., noting whether each value is larger or smaller than a threshold value), and judging which tasks are similar enough to be placed in the same second sub-group. This limitation is a process that can be performed in the human mind using evaluation, observation, and judgment, and therefore is directed to a mental process.
Step 2A Prong 2 and Step 2B:
“the computer is caused to execute a process of” -- The limitation, under its broadest reasonable interpretation, recites the use of a generic computer (a computer, and a program stored on a generic recording medium) to perform the abstract idea. The computer is recited at a high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer. Accordingly, this additional element does not integrate the judicial exception into a practical application, and does not amount to significantly more than the judicial exception (see MPEP 2106.05(f)).
Therefore, claim 4 is non-patent eligible. Claims 8 and 12 are analogous to claim 4, aside from claim type and minute differences, and thus the same rejection applies as above.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 3-5, 7-9, and 11-12 are rejected under 35 U.S.C. 103 as being unpatentable over the NPL reference “Learning Meta-Features for AutoML” by Rakotoarison et al. (referred herein as Rakotoarison) in view of the NPL reference “On Data and Algorithms: Understanding Inductive Performance” by Kalousis et al. (referred herein as Kalousis) further in view of US 20100287534A1 by Vangala et al. (referred herein as Vangala).
Regarding claim 1, Rakotoarison teaches:
executing automated machine learning (AutoML) processing on each of a plurality of tasks to acquire a plurality of pipelines for each of the plurality of tasks; ([Rakotoarison, page 7, sec. 4.1] “The considered AutoML benchmark is the OpenML Curated Classification suite 2018 (Bischl et al., 2019), including 72 binary or multi-class datasets…For AutoSkLearn, the target representation is generated from scratch, running 500 configurations per training dataset and retaining the top-20.” AND [Rakotoarison, page 1, sec. 1] “An ML pipeline consists of a data preparation stage followed by the model learning stage.” AND [Rakotoarison, page 4, sec. 3] “The descriptive features of the ψ(xi), referred to as METABU meta-features, are meant to both be cheaply computable from the basic meta-features, and define a Euclidean distance conducive to the AutoML task.”, wherein the examiner interprets the classification datasets of the AutoML benchmark to be the same as the plurality of tasks because they are both sets of data on which a machine learning problem is to be solved, and interprets the 500 AutoSkLearn configurations run per training dataset to be the same as the plurality of pipelines for each of the plurality of tasks because they are both candidate ML pipelines produced by AutoML processing for a given dataset.)
based on similarities of one or more pipelines selected based on evaluation values, among the plurality of pipelines, ([Rakotoarison, page 1-2, sec. 1] “Relevant meta-features are expected to define a reliable topology on the dataset space, such that two datasets are close iff the best hyper-parameter configurations for these datasets are close.” AND [Rakotoarison, page 4, sec. 3] “the set of hyper-parameter configurations with performance in the top-L known configuration performances (L = 20 in the experiments).”, wherein the examiner interprets the top-L configurations retained according to their performance to be the same as the one or more pipelines selected based on evaluation values because they are both the best-performing pipelines chosen according to their evaluated performance, and interprets two datasets being close when their best hyper-parameter configurations are close to be the same as similarities of the selected pipelines because they are both a measure of how alike the selected pipelines of two tasks are.)
Rakotoarison does not teach classifying the plurality of tasks into a plurality of groups, or that the classifying is further based on similarities of evaluation values of the one or more pipelines.
Kalousis teaches classifying the plurality of tasks into a plurality of groups based on similarities of one or more pipelines selected based on evaluation values, among the plurality of pipelines, and similarities of evaluation values of the one or more pipelines; ([Kalousis, page 279, sec. 3.3] “Cell (i, j) gives the ranking of the j algorithm for the i dataset.” AND [Kalousis, page 282, sec. 5-5.1] “Now we will work in the opposite direction and use the same data to discover similarities between datasets…We will use directly the EC matrix introduced in Section 3.3 to discover clusters of datasets with specific patterns of error correlations between algorithms” AND [Kalousis, page 284, sec. 5.2] “We will now try to chart the dataset space based on the relative performance of learning algorithms. To this end we will perform the same procedure as in the previous section but this time on the rankings matrix R. Applying the hierarchical clustering algorithm we discover four main clusters of datasets”, wherein the examiner interprets discovering clusters of datasets to be the same as classifying the plurality of tasks into a plurality of groups because they are both partitioning the learning problems into groups, interprets the ranking of the algorithms for each dataset to be the same as the pipelines selected based on evaluation values because they are both the learning models of a task ordered and chosen according to their evaluated performance, and interprets the patterns of error correlations between algorithms to be the same as similarities of evaluation values because they are both a comparison of the evaluated results of the learning models of each task.)
Rakotoarison and Kalousis do not teach A non-transitory computer-readable recording medium storing a task control program for causing a computer to execute a process comprising:…generating a task group by selecting one task from each of the plurality of groups based on an execution time of the AutoML processing of each of the plurality of tasks.
Vangala teaches:
A non-transitory computer-readable recording medium storing a task control program for causing a computer to execute a process comprising: ([Vangala, [0198]] “The system memory 1315 is coupled to the bus 1305 for storing information and instructions to be executed by the processing unit 1310,” AND [Vangala, [0199]] “In an embodiment, a storage device 1320, such as a magnetic or optical disk, is also coupled to the bus 1305 for storing information, including program code comprising instructions and/or data.” AND [Vangala, [0201]] “The processing unit 1310 executes one or more sequences of one or more program instructions contained in the system memory 1315.”, wherein the examiner interprets the system memory and the storage device, such as a magnetic or optical disk, storing program code comprising instructions to be executed by the processing unit to be the same as the non-transitory computer-readable recording medium storing a task control program because they are both physical storage media that store program instructions for execution, and interprets the processing unit executing the sequences of program instructions contained in the system memory to be the same as causing a computer to execute a process because they are both a computer processor carrying out stored instructions.)
generating a task group by selecting one task from each of the plurality of groups based on an execution time of the AutoML processing of each of the plurality of tasks. ([Vangala, [0038]] “In an embodiment test cases are grouped into one or more clusters 114.” AND [Vangala, [0049]] “In an embodiment a pivot test case is identified for each cluster 118…In an embodiment additional factors, e.g., test case execution time, test case type, etc., can be used to identify the pivot test case of a cluster 118.” AND [Vangala, [0050]] “In an embodiment the pivot test case of each cluster is used to generate a minimum test suite 120.”, wherein the examiner interprets the minimum test suite generated from the pivot test case of each cluster to be the same as the task group generated by selecting one task from each of the plurality of groups because they are both a reduced set formed from a single representative of each group, and interprets the test case execution time to be the same as the execution time of each of the plurality of tasks because they are both the time required to execute each member of the set being reduced.)
Rakotoarison, Kalousis, Vangala, and the instant application are analogous art because they are all directed to grouping a plurality of evaluated items, such as datasets or test cases, according to the similarity of the results obtained by executing them, in order to characterize or reduce the set of items to be evaluated.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the AutoML dataset-similarity approach based on top-performing hyper-parameter configurations disclosed by Rakotoarison to include the clusters of datasets disclosed by Kalousis. One would be motivated to do so to effectively identify the region of the dataset space to which a dataset belongs, as suggested by Kalousis ([Kalousis, page 284, sec. 5.2] “In the next section we will seek a description of areas defined by the clustering processes in terms of dataset characteristics. Our primary goal is attaining an understanding of these areas by means of the dataset characteristics. This could be used to identify the regions of the dataset space to which a dataset belongs providing an informative way to select or combine appropriate algorithm(s).”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the memory system for storing information and instructions and the “pivot test case is identified for each cluster” disclosed by Vangala. One would be motivated to do so to efficiently reduce the number of items that must be run, as suggested by Vangala ([Vangala, [0047]] “Thus, in embodiments the variance and commonality threshold values can be adjusted based on a user's needs, e.g., the threshold values can be relaxed to identify a smaller set of test cases necessary to be run in time constrained circumstances.”). Claims 5 and 9 are analogous to claim 1, aside from claim type and minute differences, and thus the same rejection applies as above.
Regarding claim 3, Rakotoarison, Kalousis, and Vangala teaches The non-transitory computer-readable recording medium according to claim 1, (see rejection of claim 1).
Vangala further teaches wherein the computer is caused to execute a process of ([Vangala, [0201]] “The processing unit 1310 executes one or more sequences of one or more program instructions contained in the system memory 1315.”, wherein the examiner interprets the processing unit executing sequences of program instructions contained in memory to be the same as a computer executing a process because they are both directed to using a computer/system to instruct/execute certain tasks.)
Kalousis further teaches further classifying the plurality of groups into a plurality of first sub-groups based on a deviation between the evaluation values in the one or more pipelines. ([Kalousis, page 278, sec. 3.1] “The relative performance of a set of algorithms on a given dataset can be depicted by their ranking based on the statistical significance of the performance differences observed from all pairwise comparisons.” AND [Kalousis, page 289, sec. 6.2] “Based on the results of dataset clustering using the rankings of the algorithms, figure 3, we can divide the datasets in two groups.” AND [Kalousis, page 290, Table 3] “Cross classification in low, high, average error correlation and equiperformance and nonequiperformance groups” AND [Kalousis, pages 289-290, sec. 6.2] “the Non-Equiperformance group contains datasets that belong either to the Low or to the High Error Correlation group, more over it contains some datasets that were not included in the High or Low Error Correlation groups.” AND [Kalousis, page 294, sec. 6.4] “The first, ω1, will contain all the datasets for which c50boost is significantly better than c50tree, (18 datasets), while the second, ω2, contain the datasets for which there is no significant difference between the two learners, (60 datasets)”, wherein the examiner interprets cross-classifying the Equiperformance and Non-Equiperformance groups, which are obtained from the ranking-based dataset clusters, into low, high and average error correlation groups, and dividing the datasets into those for which one learner is significantly better than another and those with no significant difference, to be the same as further classifying the plurality of groups into a plurality of first sub-groups because they are both dividing already-formed groups of tasks into smaller groups, and interprets the statistically significant performance differences between the learning algorithms to be the same as a deviation between the evaluation values because they are both the amount by which the evaluated results of the learning models of a task differ from one another.)
Rakotoarison, Kalousis, Vangala, and the instant application are analogous art because they are all directed to grouping a plurality of evaluated items, such as datasets or test cases, according to the similarity of the results obtained by executing them, in order to characterize or reduce the set of items to be evaluated.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the non-transitory computer-readable medium of claim 1 disclosed by Rakotoarison, Kalousis, and Vangala to include the processing unit (computer) executing instructions disclosed by Vangala. One would be motivated to do so to efficiently execute stored program instructions either as they are received or at a later time, as suggested by Vangala ([Vangala, [0204]] “The program instructions received by the computing device system 1300 may be executed by the processing unit 1310 as they are received, and/or stored in the storage device 1320 or other non-volatile storage for later execution.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the clusters of datasets and cross-classifying the Equiperformance and Non-Equiperformance groups obtained from the ranking-based dataset clusters disclosed by Kalousis. One would be motivated to do so to effectively identify the region of the dataset space to which a dataset belongs, as suggested by Kalousis ([Kalousis, page 284, sec. 5.2] “In the next section we will seek a description of areas defined by the clustering processes in terms of dataset characteristics. Our primary goal is attaining an understanding of these areas by means of the dataset characteristics. This could be used to identify the regions of the dataset space to which a dataset belongs providing an informative way to select or combine appropriate algorithm(s).”). Claims 7 and 11 are analogous to claim 3, aside from claim type and minute differences, and thus the same rejection applies as above.
Regarding claim 4, Rakotoarison, Kalousis, and Vangala teaches The non-transitory computer-readable recording medium according to claim 3, (see rejection of claim 3).
Rakotoarison further teaches based on degrees of similarity of hyper parameters of a machine learning model. ([Rakotoarison, page 1-2, sec. 1] “Relevant meta-features are expected to define a reliable topology on the dataset space, such that two datasets are close iff the best hyper-parameter configurations for these datasets are close…The target one, out-of-reach except for the datasets in the benchmark suite, represents a dataset as the distribution of the hyper-parameter configurations of A yielding the top performances for this dataset.” AND [Rakotoarison, page 4, Fig. 1] “Top configurations of datasets A, B, and C, where B, in orange (resp. C, in green) is the nearest neighbor of A w.r.t. target (resp. basic) representation.” AND [Rakotoarison, page 5, Fig. 2] “Neighbor datasets in the target space have same color in all subplots.” AND [Rakotoarison, page 7, sec. 4.1] “METABU is validated in the context of three ML algorithms: Adaboost (Freund & Schapire, 1997), RandomForest (Breiman, 2001) and SVM (Boser et al., 1992), using their scikit-learn implementation (Pedregosa et al., 2011)”, wherein the examiner interprets the closeness of the best hyper-parameter configurations of two datasets, by which a dataset’s nearest neighbor is determined in the target representation and by which neighbor datasets are assigned the same color, to be the same as degrees of similarity of hyper parameters because they are both a measure of how alike the tuned hyperparameter values of a learning model are between tasks that is used to decide which tasks belong together, and interprets Adaboost, RandomForest and SVM to be the same as a machine learning model because they are both learning algorithms whose hyperparameters are configured.)
Vangala further teaches wherein the computer is caused to execute a process of ([Vangala, [0201]] “The processing unit 1310 executes one or more sequences of one or more program instructions contained in the system memory 1315.”, wherein the examiner interprets the processing unit executing sequences of program instructions contained in memory to be the same as a computer executing a process because they are both directed to using a computer/system to instruct/execute certain tasks.)
Kalousis further teaches further classifying the plurality of first sub-groups into a plurality of second sub-groups ([Kalousis, page 294, sec. 6.4] “The first, ω1, will contain all the datasets for which c50boost is significantly better than c50tree, (18 datasets), while the second, ω2, contain the datasets for which there is no significant difference between the two learners, (60 datasets)” AND [Kalousis, page 295, sec. 6.4] “Taking a closer look at the datasets that have NCE < 0.5 we found 17. For 16 of them, c50boost and c50tree did not have a significant difference. For half of these 16 datasets, the two algorithms were significantly better than the default, ‘easy’ classification problems, while for the other half they were not significantly different than the default or they were even significantly worse, (one case).” AND [Kalousis, page 280, sec. 4] “We used an agglomerative hierarchical clustering algorithm (Duda, Hart, & Stork, 2001), together with Ward’s minimum-variance method to determine which clusters should be merged at each agglomeration step.”, wherein the examiner interprets the group ω2 of datasets with no significant difference between the two learners, which is itself formed by dividing the ranking-based dataset groups according to the deviation between the learners’ evaluated performance (see rejection of claim 3), to be the same as a first sub-group, and interprets taking the datasets of that sub-group and further dividing them into the ‘easy’ half and the remaining half to be the same as further classifying the plurality of first sub-groups into a plurality of second sub-groups because they are both a further level of division applied to an already-formed sub-group of tasks to yield still smaller groups.)
Rakotoarison, Kalousis, Vangala, and the instant application are analogous art because they are all directed to grouping a plurality of evaluated items, such as datasets or test cases, according to the similarity of the results obtained by executing them, in order to characterize or reduce the set of items to be evaluated.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the non-transitory computer-readable medium of claim 3 disclosed by Rakotoarison, Kalousis, and Vangala to include the comparison of datasets according to whether “the best hyper-parameter configurations for these datasets are close” disclosed by Rakotoarison. One would be motivated to do so to efficiently select promising hyper-parameter configurations for a dataset from its most similar datasets, as suggested by Rakotoarison ([Rakotoarison, page 2, sec. 1] “Such a topology would support an inexpensive and efficient AutoML strategy: selecting the best hyper-parameter configurations of the nearest neighbor(s) of the current dataset.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the “processing unit 1310 executes one or more sequences of one or more program instructions” disclosed by Vangala. One would be motivated to do so to efficiently execute stored program instructions either as they are received or at a later time, as suggested by Vangala ([Vangala, [0204]] “The program instructions received by the computing device system 1300 may be executed by the processing unit 1310 as they are received, and/or stored in the storage device 1320 or other non-volatile storage for later execution.”).
It would have also been obvious to a person of ordinary skill in the art before the effective filing date of the invention to include the further division of an already-formed sub-group of datasets, i.e., the “closer look at the datasets” within a group, disclosed by Kalousis. One would be motivated to do so to effectively identify the region of the dataset space to which a dataset belongs, as suggested by Kalousis ([Kalousis, page 284, sec. 5.2] “In the next section we will seek a description of areas defined by the clustering processes in terms of dataset characteristics. Our primary goal is attaining an understanding of these areas by means of the dataset characteristics. This could be used to identify the regions of the dataset space to which a dataset belongs providing an informative way to select or combine appropriate algorithm(s).”). Claims 8 and 12 are analogous to claim 4, aside from claim type and minute differences, and thus the same rejection applies as above.
Claims 2, 6, and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Rakotoarison, Kalousis, Vangala, and further in view of the NPL reference “Quantifying the Impact of Input Data Sets on Program Behavior and its Applications” by Eeckhout et al. (referred herein as Eeckhout).
Regarding claim 2, Rakotoarison, Kalousis, and Vangala teaches The non-transitory computer-readable recording medium according to claim 1, (see rejection of claim 1).
Rakotoarison, Kalousis, and Vangala do not teach wherein the one task is a task that has a smallest total value of the execution times in a group to which the one task belongs, among the plurality of groups.
Eeckhout teaches wherein the one task is a task that has a smallest total value of the execution times in a group to which the one task belongs, among the plurality of groups. ([Eeckhout, page 2, sec. 1] “Indeed, strong clustering suggests that a single or only a few input sets should be selected to be representative for the cluster. This will reduce the total simulation time significantly for two reasons: (i) the total number of benchmark-input pairs is reduced; and (ii) we can select the benchmark-input pair with the smallest dynamic instruction count among all the pairs in the cluster.” AND [Eeckhout, pages 25-26, sec. 5.1] “The representative for each cluster was chosen by taking the program-input pair with the minimal dynamic instruction count that is as close as possible to the center of the cluster it belongs to.” AND [Eeckhout, page 23, sec. 5.1] “Indeed, the total dynamic instruction count of all these program-input pairs together exceeds 593 billions of instructions, or 23 days of simulation”, wherein the examiner interprets the benchmark-input pair with the smallest dynamic instruction count among all the pairs in the cluster to be the same as the one task that has a smallest total value of the execution times in the group to which it belongs because they are both the member of a group that requires the least total execution to run, the dynamic instruction count being the total number of instructions executed by that member and directly determining its simulation time, and interprets the cluster to be the same as the group because they are both a set of similar members from which one representative is selected.)
Rakotoarison, Kalousis, Vangala, Eeckhout, and the instant application are analogous art because they are all directed to grouping a plurality of evaluated items, such as datasets, test cases or benchmark-input pairs, according to the similarity of the results obtained by executing them, in order to characterize or reduce the set of items to be evaluated.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the non-transitory computer-readable medium of claim 1 disclosed by Rakotoarison, Kalousis, and Vangala to include the “smallest dynamic instruction count among all the pairs in the cluster” disclosed by Eeckhout. One would be motivated to do so to efficiently reduce the total time needed to evaluate the selected representatives, as suggested by Eeckhout ([Eeckhout, page 2, sec. 1] “Indeed, strong clustering suggests that a single or only a few input sets should be selected to be representative for the cluster. This will reduce the total simulation time significantly for two reasons: (i) the total number of benchmark-input pairs is reduced; and (ii) we can select the benchmark-input pair with the smallest dynamic instruction count among all the pairs in the cluster.”). Claims 6 and 10 are analogous to claim 2, aside from claim type and minute differences, and thus the same rejection applies as above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DEVAN KAPOOR whose telephone number is (703)756-1434. The examiner can normally be reached Monday - Friday: 9:00AM - 5:00 PM EST (times may vary).
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, David Yi can be reached at (571) 270-7519. 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.
/DEVAN KAPOOR/Examiner, Art Unit 2126
/DAVID YI/Supervisory Patent Examiner, Art Unit 2126
Prosecution Insights