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 .
The following action is in response to the amendment and remarks of 01/08/2026.
By the amendment, claims 1, 13 and 19 have been amended.
Claims 1-20 are pending and have been considered below.
Response to Arguments
The 35 USC 101 rejection of claims 1-20 have been withdrawn in light of the claims amendment and corresponding remarks.
The 35 USC 102 rejection of claims 1-2, 7, 9-14 and 19-20 by SABERIAN and the 35 USC 103 rejection of claims 3-6, 8 and 15-18 over SABERIAN in view of SZOKE-SIESWERDA have been withdrawn in light of the amendment and corresponding remarks. The Examiner notes that while SABERIAN discloses enabling a boosting algorithm (¶30) distributed among processing units (¶12) for use in a PLC system having artificial intelligence capability (¶56, ¶30), the Examiner agrees with Applicant that neither SABERIAN nor SZOKE-SIESWERDA explicitly disclose the use of NPUs as required by the amended claims.
However, on further search and consideration, the prior art of CHOI (US 2021/0279482 A1) has been found to meet the deficiencies of SABERIAN and SZOKE-SIESWERDA as presented in the new grounds of rejection below.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-2, 7, 9-14 and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over SABERIAN, US 2017/0039456 A1 (previously presented) in view of CHOI, US 2021/0279482 A1 effective filing of 03/05/2020 (“CHOI”).
Regarding claim 1, SABERIAN discloses a computer-implemented method of distributed neural network boosting (¶8, ¶12: distributed neural network), the method performed by a neural network system and comprising:
through operating at least one processor:
providing a boosting algorithm (¶30: boosting neural network) that distributes a model among a plurality of processing units (¶11-12: distributed computing of portions of a neural network for scheduling processing jobs, ex. ¶26: using plural processors) each being a weak learner of multiple weak learners (¶30: CNN architectures comprising weak learners) that can perform computations independent from one another yet process data concurrently (¶30: individual CNNs operating in seclusion from one another), wherein each processing unit is designed for use in a programmable logic system and has artificial intelligence capability (¶26: plural processors, ¶11-12: distributed computing of portions of neural network for scheduling jobs, ¶57: processors may comprise programmable logic controller devices);
enabling a distributed ensemble learning (¶30: form an ensemble of neural networks) which enables the boosting algorithm to enable the PLC to use more than one of the processing units to scale an application (¶12: large scale distributed networks implementing schedulers for executing large processing jobs is enabled by ¶30: the ensemble boosting neural networks operating on and used by ¶12: at least one of the plurality of processors of the ¶57: PLC);
training the multiple weak learners using the boosting algorithm (¶45: trained networks), wherein the multiple weak learners are machine learning models that do not capture an entire data distribution (¶30: individual weak learner CNNs operating in seclusion from one another)) and are purposefully designed to predict with a lower accuracy (¶30: slightly correlated with accuracy); and
using the multiple weak learners to vote for a final hypothesis based on a feed forward computation of neural networks (¶48: aggregate outputs of active networks through forward passes to compute a boosting predictor).
SABERIAN fails to disclose wherein the processing units designed for use in a PLC system are Neural Processing Units each being an edge device.
CHOI discloses methods for processing an image using machine learning (¶4). In particular, CHOI discloses enabling boosting models (¶35) utilizing edge device processing circuity that includes neural network units and programmable logic units (¶44, ¶73, ¶83, ¶104). Accordingly, it would have been obvious to one having ordinary skill in the art and the teachings of SABERIAN and CHOI before them to combine the edge device neural processing units for use a PLC system of CHOI with the processing units for use in a PLC system of SABERIAN yielding the predictable result of enabling the distributed ensemble learning of the combination of SABIERAN and CHOI by enabling the boosting algorithm utilizing a PLC and one or more edge device NPUs. One would have been motivated to make this combination in order to utilize known circuity for accomplishing the implementation goals of increased reliability of detected objects, as suggested by CHOI (¶4, ¶26-27, ¶44).
Regarding claim 2, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses:
using neural networks as the multiple weak learners (¶30).
Regarding claim 7, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses:
using the boosting algorithm directly for a typical regression task or a classification task (¶25).
Regarding claim 9 SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses wherein the boosting algorithm guarantees a reduction in variance without increasing a bias thus making the model more generalizable (¶45).
Regarding claim 10, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses wherein the boosting algorithm combines multiple distributed neural network models to create a more complex model without reaching a resource limitation (¶12, ¶45).
Regarding claim 11, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses:
combining all outputs to allow the model to expand a Vapnik-Chervonenkis (VC) dimension effectively covering a larger underlying distribution of training data (¶44-45: reduction in complexity while increasing training space).
Regarding claim 12, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses:
with the boosting algorithm training, each weak learner of the multiple weak learners focuses on a resampled subset of a dataset (¶43).
Regarding claims 13-14, claims 13-14 recite limitations similar to claims 1-2, respectively, and are similarly rejected.
Regarding claims 19-20, claims 19-20 recite limitations similar to claims 1-2, respectively, and are similarly rejected.
Claims 3-6, 8 and 15-18 are rejected under 35 U.S.C. 103 as being unpatentable over SABERIAN in view of CHOI and in further view of SZOKE-SIESWERDA, US 2020/0342240 A1 (previously presented).
Regarding claim 3, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses:
applying the boosting algorithm to image detection (¶25: structural feature extraction with visible objects, ¶29-31) and using CNNs as a weak learner (¶30) to intentionally rendering the model weak (¶30).
SABERIAN and CHOI fail to disclose using a single shot detector (SSD) as a weak learner while at least two hyperparameters are used to intentionally render the model weak.
SZOKE-SIESWERDA discloses methods for performing image detection using neural networks (¶12, ¶77). In particular, SZOKE-SIESWERDA discloses using a single shot detector in place of a CNN (¶83) when performing object detection (¶83) while at least two hyperparameters can be used to intentionally change model complexity (¶80) in known CNN architectures for feature extraction (¶76). Therefore it would have been obvious to one having ordinary skill in the art and the teachings of SABERIAN, CHOI and SZOKE-SIESWERDA before them before the effective filing of the claimed invention to combine the use of a single shot detector as a substitute for a CNN in an image object detection model while hyperparameters are used to intentionally change model complexity, as taught by SZOKE-SIESWERDA, with the use of CNNs as weak learners rendering the model weak when applying the boosting algorithm of SABERIAN and CHOI to image object detection tasks. One would have been motivated to make this combination to simply substitute, using known methods, known architecture used for object detection and potentially yielding lower latency, as suggested by SZOKE-SIESWERDA (¶76, ¶83-84).
Regarding claim 4, SABERIAN, CHOI and SZOKE-SIESWERDA disclose the method of claim 3, and SZOKE-SIESWERDA discloses wherein the at least two hyperparameters include a width multiplier that thins the neural network system at each layer (¶80-81).
Regarding claim 5, SABERIAN, CHOI and SZOKE-SIESWERDA disclose the method of claim 4, and SZOKE-SIESWERDA discloses wherein an accuracy of the model and speed is adjusted with the at least two hyperparameters (¶80-81).
Regarding claim 6, SABERIAN, CHOI and SZOKE-SIESWERDA disclose the method of claim 5, and SZOKE-SIESWERDA discloses wherein each model of the multiple weak learners of SABERIAN, CHOI and SZOKE-SIESWERDA returns a list of output bounding boxes and their respective classes (¶84).
Regarding claim 8, SABERIAN and CHOI disclose the method of claim 1, and SABERIAN further discloses :
using the boosting algorithm for an image detection task (¶25: structural feature extraction with visible objects, ¶29-31), wherein each model of the multiple weak CNN learners returns their respective classes (¶46).
SABERIAN and CHOI fail to disclose wherein each model of the multiple weak learners returns a list of output bounding boxes; grouping together all the output bounding boxes and all the classes into a set such that the set contains many low-confidence predictions and duplicates; discarding the many low-confidence predictions by using a threshold; and applying non-maximum suppression to reduce the duplicates.
SZOKE-SIESWERDA discloses methods for performing image detection using neural networks (¶12, ¶77). In particular, SZOKE-SIESWERDA discloses when performing feature extraction for an image detection task (79-83), returning a list of output bounding boxes (¶83-84); grouping together all the output bounding boxes and all the classes into a set such that the set contains many low-confidence predictions and duplicates (¶85); discarding the many low-confidence predictions by using a threshold (¶85); and applying non-maximum suppression to reduce the duplicates (¶86). Therefore it would have been obvious to one having ordinary skill in the art and the teachings of SABERIAN, CHOI and SZOKE-SIESWERDA before them before the effective filing of the claimed invention to combine the returning a list of output bounding boxes, grouping together all the output bounding boxes and all the classes into a set such that the set contains many low-confidence predictions and duplicates, discarding the many low-confidence predictions by using a threshold, and applying non-maximum suppression to reduce the duplicate when performing an image detection task, as taught by SZOKE-SIESWERDA, with the performing an image detection task of SABERIAN and CHOI wherein the multiple weak learner CNNs of output their respective classes. One would have been motivated to make this combination to use known architecture used for object detection and potentially yielding lower latency, as suggested by SZOKE-SIESWERDA (¶76,¶83-84).
Regarding claims 15-17, claims 15-17 recite limitations similar to claims 3-5 and are similarly rejected.
Regarding claim 18, SABERIAN, CHOI and SZOKE-SIESWERDA disclose the system of claim 15, and SZOKE-SIESWERDA discloses wherein each model of the multiple weak learners of SABERIAN and SZOKE-SIESWERDA returns a list of output bounding boxes and their respective classes (¶84).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Ryan et al.
US 20210089927 A9
UNSUPERVISED OUTLIER DETECTION IN TIME-SERIES DATA
Parr et al.
US 20230120932 A1
AUTONOMOUS DATA COLLECTION AND SYSTEM CONTROL FOR MATERIAL RECOVERY FACILITIES
Vaghela, Vimal B., Amit Ganatra, and Amit Thakkar. "Boost a weak learner to a strong learner using ensemble system approach." 2009 ieee international advance computing conference. IEEE, 2009.
Pias, Marcelo, Silvia Botelho, and Paulo Drews. "Perfect storm: DSAs embrace deep learning for GPU-based computer vision." 2019 32nd SIBGRAPI Conference on Graphics, Patterns and Images Tutorials (SIBGRAPI-T). IEEE, 2019.
Rincy, Thomas N., and Roopam Gupta. "Ensemble learning techniques and its efficiency in machine learning: A survey." 2nd international conference on data, engineering and applications (IDEA). IEEE, 2020.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 nonprovisional extension fee (37 CFR 1.17(a)) 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 ANDREW L TANK whose telephone number is (571)270-1692. The examiner can normally be reached Monday-Thursday 9a-6p.
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, Matthew Ell can be reached at 571-270-3264. 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.
/ANDREW L TANK/Primary Examiner, Art Unit 2141
Prosecution Insights