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
The action is in response to claims dated 10/28/2022
Claims pending in the case: 1-25
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.
Claim(s) 1-7, 9-17, 19-25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yao (Automatic Weakly Supervised Object Detection From High Spatial Resolution Remote Sensing Images via Dynamic Curriculum Learning) in view of Arik (CN 114424212).
Regarding Claim 1, Yao teaches, A method, comprising:
accessing a deep neural network (DNN) that has been trained to receive an input and to output a class of the input and a confidence score indicating a likelihood of the input falling into the class (Yao: Pg. 677 col 2 [1]: baseline DNN with classification and detection score);
inputting calibration samples into the DNN, the DNN outputting classes of the calibration samples, the calibration samples associated with ground-truth labels indicating ground-truth classes of the calibration samples (Yao: Pg. 679 col 1 section B [1-2]: collaborative learning strategy using samples labeled with ground-truth); and
training a calibration function associated with the DNN by optimizing a value of a first learnable parameter of the calibration function and a value of a second learnable parameter of the calibration function based on the classes of the calibration samples and the ground-truth classes of the calibration samples (Yao: Pg. 679 col 2 section B-2 [2]: a focal loss function parameter 1) is designed with a weighting function (parameter2); Pg. 680 col 1 Algorithm 1: dynamic learning),
wherein … after the training … determine a new confidence score indicating a new likelihood of the input falling into the class (Yao: Pg. 680 col 1 section B-3: a new detection model is learned (new probability score), Pg. 679 col 1 [1], Pg. 683 col 2 section D [2]: get image difficulty score which may be confidence-based);
Although Yao does not recite determining a confidence score, Yao teaches, calibrating a function to determine a difficulty score which may be confidence based. Therefore the difficulty score is a measure of the confidence and hence the examiner finds that the limitations as claimed would be obvious over the teachings in Yao;
Nonetheless Arik teaches, determine a new confidence score (Arik: Pg. 2, [2-3], Pg. 3 [2], Pg. 6 [1], Pg. 7 [2]: a calibration function trained to determine new confidence score);
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Yao and Arik because the combination would enable using a function for determining a confidence score. One of ordinary skill in the art would have been motivated to combine the teachings because the combination would enable using combined training to generate an accurate confidence score (See Arik Pg. 2 [1]).
Regarding claim 2, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein training the calibration function comprises: optimizing the value of the first learnable parameter and the value of the second learnable parameter by minimizing a loss between the classes of the calibration samples and the ground-truth classes of the calibration samples (Yao: Pg. 679 col 1 section B [1-2]: collaborative learning strategy using samples labeled with ground-truth; Pg. 679 col 2 section B-2 [2]: a focal loss function parameter 1) is designed with a weighting function (parameter2); Pg. 680 col 1 Algorithm 1: dynamic learning) (Arik: Pg. 6 [1], Pg. 7 [2]: minimize loss function).
Regarding claim 3, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein the value of the first learnable parameter or the value of the second learnable parameter is above zero (Yao: Pg. 679 col 1 section B [1-2]: collaborative learning updating parameters; Pg. 681 col 1 [2]: positive values) (Arik: Pg. 7 [2]: update parameters). Training parameters may be positive. It is noted here that the parameter being positive is non-functional and is not functionally involved in the steps recited. Thus, this descriptive material will not distinguish the claimed invention from the prior art in terms of patentability.
Regarding claim 4, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein a value of the new confidence score decreases as the value of the first learnable parameter or the value of the second learnable parameter increases (Arik: Pg. 6 [4]-Pg. 7 [1]: confidence score based on distance – decreases with distance).
Regarding claim 5, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein: associating the calibration function with the DNN comprises associating the calibration (Yao: Pg. 679 col 1 section B [1-2]: collaborative learning) (Arik: Pg. 2, [2-3]: combined training)
Regarding claim 6, Yao and Arik teach the invention as claimed in claim 5 above and,
wherein the calibration function is a function of an entropy of a vector generated by the one or more of hidden layers (Yao: Pg. 676 col 1 [2], Pg. 678 col 2 [2] - Pg. 679 col 1 [1], Pg. 683 Table III. Section D [1]: entropy based)
Regarding claim 7, Yao and Arik teach the invention as claimed in claim 5 above and,
wherein the output layer includes a softmax function that determines the likelihood of the input falling into the class (Yao Pg. 978 col 1 [1]: softmax to assign to a class).
Regarding claim 9, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein an accuracy of the DNN before training the calibration function is the same as an accuracy of the DNN after training the calibration function (Arik: Pg. 2, [2-3], Pg. 3 [2], Pg. 6 [1], Pg. 7 [2]: a calibration function separate from the model – model accuracy not affected).
Regarding claim 10, Yao and Arik teach the invention as claimed in claim 1 above and,
wherein the DNN has been trained by:
inputting one or more training samples into the DNN, the DNN outputting classes of the one or more training samples, the one or more training samples associated with ground-truth labels indicating ground-truth classes of the one or more training samples (Yao: Pg. 679 col 1 section B [1-2]: collaborative learning strategy using samples labeled with ground-truth) (Arik: Pg. 3 [2], Pg. 6 [1]:input data); and
optimizing values of internal parameters of the DNN based on the classes of the one or more training samples and the ground-truth classes of the one or more training samples (Yao: Pg. 679 col 2 section B-2 [2]: a focal loss function parameter 1) is designed with a weighting function (parameter2); Pg. 680 col 1 Algorithm 1: dynamic learning) (Arik: Pg. 2, [2-3], Pg. 3 [2], Pg. 6 [1], Pg. 7 [2]: a calibration function trained to determine new confidence score),
wherein the one or more training samples are different from the calibration samples (Yao: Pg. 679 col 1 section B [1]: use different sample sets based on difficulty score) (Arik: Pg. 5 [3]: a subset of the training samples).
Regarding Claim(s) 11-17, 19-20, this/these claim(s) is/are similar in scope as claim(s) 1-7, 9-10 respectively. Therefore, this/these claim(s) is/are rejected under the same rationale.
Regarding Claim(s) 21-25, this/these claim(s) is/are similar in scope as claim(s) 1-3, 5 and 10 respectively. Therefore, this/these claim(s) is/are rejected under the same rationale
Claim(s) 8, 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yao (Automatic Weakly Supervised Object Detection From High Spatial Resolution Remote Sensing Images via Dynamic Curriculum Learning) and Arik (CN 114424212) in view of Gautham (US 20240095539).
Regarding claim 8, Yao and Arik teach the invention as claimed in claim 1 above and,
further comprising: verifying an accuracy of the DNN based on the classes of the calibration samples and the ground-truth classes of the calibration samples, … (Yao: Pg. 680 col 1 section B-3: difficulty score) (Arik: Pg. 3 [2]: accuracy based on ground truth distance);
Yao and Arik teaches accuracy and difficulty score but does not specifically recite, the accuracy indicated by a ratio of a number of one or more calibration samples that the DNN correctly classified to a total number of the calibration samples; This however is well known in the art;
Nonetheless, Gautham teaches, accuracy indicated by a ratio of a number of one or more calibration samples that the DNN correctly classified to a total number of the calibration samples (Gautham: [50] algorithm: “accuracy function, defined by the ratio of correctly classified samples to the total samples for the given local model”);
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Yao, Arik and Gautham because the combination would enable using a ratio of correctly classified to the total as a measure of accuracy of the model. One of ordinary skill in the art would have been motivated to combine the teachings because the combination would enable using an accuracy measure common in the art.
Regarding Claim(s) 18, this/these claim(s) is/are similar in scope as claim(s) 8. Therefore, this/these claim(s) is/are rejected under the same rationale.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure in attached 892.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MANDRITA BRAHMACHARI whose telephone number is (571)272-9735. The examiner can normally be reached Monday to Friday, 11 am to 8 pm EST.
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, Tamara Kyle can be reached at 571 272 4241. 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.
/Mandrita Brahmachari/Primary Examiner, Art Unit 2144