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 .
Specification
The specification has been checked, but not to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Drawings
The applicant’s submitted drawings appear to be acceptable for examination purposes. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the drawings.
Priority
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Information Disclosure Statement
The listing of references in the specification is not a proper information disclosure statement. 37 CFR 1.98(b) requires a list of all patents, publications, or other information submitted for consideration by the Office, and MPEP § 609.04(a) states, "the list may not be incorporated into the specification but must be submitted in a separate paper." Therefore, unless the references have been cited by the examiner on form PTO-892, they have not been considered.
The information disclosure statement filed 1 March 2023 fails to comply with 37 CFR 1.98(a)(2), which requires a legible copy of each cited foreign patent document; each non-patent literature publication or that portion which caused it to be listed; and all other information or that portion which caused it to be listed. It has been placed in the application file, but the information referred to therein has not been considered. Examiner’s Note: a copy appears to have been filed that may be related to the cited reference to MCCLOSKEY, MICHAEL, but the copy is illegible and only appears to include a cover page, while the citation is to pgs. 109-165.
Claim Objections
Claim 2 is objected to because of the following informalities: “augmenting multiples correlated views” appears as though it should be “augmenting multiple correlated views” or similar. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-8 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Regarding claim 1, the phrase "such as" renders the claim indefinite because it is unclear whether the limitations following the phrase are part of the claimed invention. See MPEP § 2173.05(d).
Claims 2-8 depend upon claim 1, and thus include the aforementioned limitation(s).
Claim 4 also recites the limitation "said correlated views” in lines 2 and 3. There is insufficient antecedent basis for this limitation in the claim. Examiner’s Note: it appears that claim 4 is intended to depend from claim 2.
Claim 6 also recites the limitations “the correlations of 12-normalized output activations of the projection head” and “the correlations of 12-normalized output activations of the classifier projection” in lines 2-4. There is insufficient antecedent basis for these limitations in the claim. Additionally, the intended scope of the claim is not clear because it is not clear what is meant by 12-normalized output activations. The examiner has assumed, for the purposes of examination, that these are L2 normalized output activations.
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.
Claim 7 is rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. Specifically, according to the description given in the specification, in paragraph [0008], the broadest reasonable interpretation of a “computer-readable medium” covers transitory propagating signals, which are non-statutory. To overcome this rejection, applicant should insert --non-transitory-- before “computer-readable medium.” Such an amendment is not considered new matter. See the "Subject Matter Eligibility of Computer Readable Media" memo dated January 26, 2010 (OG Cite: 1351 OG 212; OG Date: 23 Feb 2010).
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.
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.
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.
Claim(s) 1-3 and 5 is/are rejected under 35 U.S.C. 103 as being unpatentable over Cha et al. (Co2L: Contrastive Continual Learning, June 2021, pgs. 1-14 – cited in an IDS) in view of Arani et al. (Learning Fast, Learning Slow: A General Continual Learning Method Based on Complementary Learning System, May 2022, pgs. 1-22 – cited in an IDS).
As per claim 1, Cha teaches a computer-implemented method for learning of an artificial neural network on an input of a continual stream of tasks, the method comprising a continual learning model [a Co2L model framework, including multiple models (pg. 2, fig. 1; etc.) for contrastive continual learning (pg. 1, abstract; etc.), where the models include a convolutional neural network (pg. 11, section A.2; etc.); so the Co2L framework is the continual learning model] comprising the steps of:
maintaining a fixed-size memory buffer using reservoir sampling for storing data distributions of previous tasks [samples are drawn from a memory buffer (pg. 2, fig. 1; etc.) using a class balanced sampling strategy from the buffer (pg. 6, section 5.1; etc.) with a predetermined (fixed) memory size (pg. 7, section 5.2; etc.); where the sampling from the memory buffer is the reservoir sampling];
providing the network with an encoder, a linear classifier, a classifier projection with a multi-layer perceptron, and a projection head [the model includes an encoder and projection map sequentially (pg. 5, section 4.3; etc.) with a downstream classifier (pgs. 6-7, section 5.2; etc.) and a 2-layer projection MLP producing a projection map in the projection space (projection head) (pg. 11, section A.2; etc.)];
implicitly regularizing the continual learning model by learning generalizable features through an auxiliary task such as supervised contrastive learning [auxiliary samples (an auxiliary task) are used to train the classifier (pg. 7, section 5.3; etc.) using self-supervised contrastive learning (pg. 1, abstract, etc.) which regularizes using preserved (past) knowledge (pg. 1, section 1; etc.); providing implicit regularization];
explicitly regularizing the learning model by:
calculating a moving average of parameters of the continual learning model [the classification model results are averaged over a set number of trials (pg. 7, table 1 and section 5.2; pg. 8, tables 2-3; etc.) and selecting hyperparameters that achieve the highest final accuracy averaged over both settings (pg. 11, section A.3; etc.)];
using predictions of said moving average for regularizing the continual learning model in a function space of the linear classifier and in a function space of the projection head [a loss is used that is the difference between MSE of the embedding space and the projection space (pg. 13, section B), which is regularizing the model in a function space of the linear classifier (embedding space) and the projection head (projection space)]; and
aligning geometric structures within a unit hypersphere of the linear classifier with a unit hypersphere of the projection head [Since we learn representations continually in contrastive learning schemes, where similarity is defined on a unit d-dimensional Euclidean sphere, regulating the relation drifts in the projection space can be more effective to preserve learned representations than other alternatives (pg. 13, section B; etc.); where the regulation is the aligning geometric structures within the unit d-dimensional Euclidean sphere (hypersphere)].
While Cha teaches using a moving average of results to regularize the continual learning model (see above), it has not been relied upon for teaching calculating an exponentially moving average of parameters of the continual learning model, and using predictions of said exponentially moving average.
Arani teaches calculating an exponentially moving average of parameters of the continual learning model, and using predictions of said exponentially moving average for regularizing the continual learning model [the semantic memories are maintained by taking the exponential moving average of the working model’s weights to consolidate information across tasks and using them as a teacher for semi-supervised learning (pg. 2, section 1; pg. 3, section 3.2; etc.)].
Cha and Arani are analogous art, as they are within the same field of endeavor, namely continual learning.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to calculate and utilize the exponentially moving average of parameters of the continual learning model, and using predictions of said exponentially moving average, as taught by Arani, for the semi supervised learning and regularization in a function space of the linear classifier and in a function space of the projection head, in the system taught by Cha.
Arani provides motivation as [the semantic memories are maintained by taking the exponential moving average of the working model’s weights to consolidate information across the tasks with varying time windows and frequencies (pg. 2, section 1; etc.), which serves as an efficient method for aggregating the weights of the model for semi-supervised learning (as used in Cha, above) (pg. 3, section 3.2; etc.)].
As per claim 2, Cha/Arani teaches wherein the step of learning generalizable features through an auxiliary task comprises the steps of:
augmenting multiples correlated views of an input sample [the batches used by the model are composed of multiple, independently augmented views of N samples (Cha: pg. 4, section 4; pg. 5, section 4.3; etc.)]; and
propagating said correlated views forward through the encoder and a projection head of the network [the model includes an encoder and projection map sequentially (Cha: pg. 5, section 4.3; etc.) with a downstream classifier (Cha: pgs. 6-7, section 5.2; etc.) and a 2-layer projection MLP producing a projection map in the projection space (projection head) (Cha: pg. 11, section A.2; etc.) where the batches used by the model are composed of multiple, independently augmented views of N samples (Cha: pg. 4, section 4; pg. 5, section 4.3; etc.)].
As per claim 3, Cha/Arani teaches further comprising the step of creating positive and negative embedding pairs of input samples using label information, wherein input samples belonging to a same class of an anchor are labelled as positives, and wherein input samples belonging to a different class than the class of the anchor are labelled as negatives [current task samples are used as anchors and positive samples of the same class (Cha: pg. 4, fig. 2, etc.), while past task samples are used as negative samples (Cha: pg. 4, section 4.1; etc.); where the current/past tasks are the different classes of the samples (see also: Cha: pg. 4, fig. 2; etc.)].
As per claim 5, Cha/Arani teaches further comprising the step of using a mapping function for connecting geometric relationships between points of the unit hypersphere of the classifier and points of the unit hypersphere of the projection head [logit matching can be used including a feature map which maps augmented batch to an unnormalized d-dimensional Euclidean sphere (hypersphere) and a difference calculated between the embedding loss and projection loss as a choice of representation maps to be matched, one defined on embedding space and the other defined on projection space (Cha: pg. 13, section B; etc.); which is the mapping function connecting geometric relationships between points of the unit hyperspheres (d-dimensional Euclidean spheres) of the classifier/encoder and the projection head].
Claim(s) 4 is/are rejected under 35 U.S.C. 103 as being unpatentable over Cha and Arani as applied to claim 1 above, and further in view of Lovchinsky (US 2023/0306982).
As per claim 4, Cha/Arani teaches the method according to claim 1, as described above.
While Cha/Arani also teaches an IRD loss that quantifies the discrepancy between the instance-wise similarities of the current representation and the past representation (Cha: pg. 5, section 4.2; etc.); it has not been relied upon for teaching further comprising the step of learning visual representations by maximizing a cosine similarity between positive pairs of said correlated views while simultaneously minimizing a cosine similarity between negative pairs of said correlated views.
Lovchinsky teaches learning visual representations by maximizing a cosine similarity between positive pairs of said correlated views while simultaneously minimizing a cosine similarity between negative pairs of said correlated views [The similarity may be quantified using cosine similarity and the loss function may be configured such that the optimization maximizes the cosine similarity for positive pairs and minimizes the cosine similarity for negative pairs. (para. 0156, etc.); for the positive and negative pairs of the correlated views in Cha/Arani, above].
Cha/Arani and Lovchinsky are analogous art, as they are within the same field of endeavor, namely contrastive learning models/methods.
It would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to calculate similarity and maximize/minimize the cosine similarities of positive/negative pairs, as taught by Lovchinsky, for the optimization with positive/negative pairs in the system taught by Cha/Arani.
Lovchinsky provides motivation as [by maximizing and minimizing the different cosine similarities, the optimization process may be configured in such a way that embeddings corresponding to samples from the same source are made as similar as possible, while embeddings corresponding to samples from different sources are optimized to be as different as possible (para. 0155, etc.)].
Claim(s) 6-8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Cha and Arani as applied to claim 1 above, and further in view of well-known practices in the art.
As per claim 6, Cha/Arani teaches further comprising the step of regularizing the output activations of the classifier projection by capturing mean element-wise squared differences in the correlations of normalized output activations of the projection head and the correlations of normalized output activations of the classifier projection [the element-wise mean squared errors of the embedding space and projection space are calculated as the choice of representation maps to be matched (Cha: pg. 13, section B; etc.)].
While Cha/Arani teaches capturing MSE of normalized output activations of the projection head and classifier projection (see above), it has not been relied upon for teaching 12-normalized output activation (interpreted as L2-normalized – see above).
However, the examiner takes official that L2-normalization of output activation is old and well-known within the art of normalizing the activation outputs to a Euclidean norm. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to perform L2-normalization of the output activations in the system/method taught by Cha/Arani, to achieve the predictable result of producing unit regularized outputs and improving the model generalization.
As per claim 7, Cha/Arani the computer-implemented method according to claim 1, as described above.
Cha/Arani has not been relied upon for teaching a computer-readable medium provided with a computer program, wherein the computer program is loaded and executed by a computer, the computer program causes the computer to carry out the steps of [the method].
However, the examiner takes official notice that implementing a computer-implemented method via computer program instructions stored in a computer-readable medium and execute by one or more processors is old and well-known within the art, as a means of providing a physical implementation to run the method. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to implement the method of Cha/Arani using a computer-readable medium provided with a computer program, wherein the computer program is loaded and executed by a computer, the computer program causes the computer to carry out the steps of [the method], to achieve the predictable result of providing a way to actually physically run the method.
As per claim 8, Cha/Arani the computer-implemented method according to claim 1, as described above.
Cha/Arani has not been relied upon for teaching an autonomous vehicle comprising a data processing system loaded with a computer program, wherein the program is arranged for causing the data processing system to carry out the steps of the computer-implemented method according to claim 1 for enabling the autonomous vehicle to continually adapt and acquire knowledge from an environment surrounding the autonomous vehicle.
However, the examiner takes official notice that implemented continual learning in an autonomous vehicle comprising a data processing system loaded with a computer program, wherein the program is arranged for causing the data processing system to perform the continual learning is old and well-known within the art for utilizing the continually updated data received by the vehicle to improve the learning system. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date of the claimed invention, to implement the method taught by Cha/Arani in in an autonomous vehicle comprising a data processing system loaded with a computer program, wherein the program is arranged for causing the data processing system to perform the method, to achieve the predictable result of allowing the continual learning to learn from the data received by the vehicle continuously, in order to improve the model and the systems of the autonomous vehicle.
Conclusion
The following is a summary of the treatment and status of all claims in the application as recommended by M.P.E.P. 707.07(i): claims 1-8 are rejected.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Li et al. (CoMatch: Semi-supervised Learning with Contrastive Graph Regularization, Mar 2021, pgs. 1-11) – discloses a semi-supervised contrastive learning system/method that jointly learns two representations of the training data, their class probabilities and low-dimensional embeddings.
Pan et al. (Click-through Rate Prediction with Auto-Quantized Contrastive Learning, Sept 2021, pgs. 1-14) – discloses a contrastive learning system/method for CTR prediction using an Auto-Quantized Contrastive Learning (AQCL) loss to regularize the model.
Wang et al. (Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hypersphere, Aug 2022, pgs. 1-41 – cited in an IDS) – discloses a contrastive learning system/method utilizing alignment (closeness) of features from positive pairs, and uniformity of the induced distribution of the normalized features on a hypersphere.
The examiner requests, in response to this Office action, that support be shown for language added to any original claims on amendment and any new claims. That is, indicate support for newly added claim language by specifically pointing to page(s) and line number(s) in the specification and/or drawing figure(s). This will assist the examiner in prosecuting the application.
When responding to this office action, Applicant is advised to clearly point out the patentable novelty which he or she thinks the claims present, in view of the state of the art disclosed by the references cited or the objections made. He or she must also show how the amendments avoid such references or objections. See 37 CFR 1.111(c).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GEORGE GIROUX whose telephone number is (571)272-9769. The examiner can normally be reached M-F 10am-6pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Omar Fernandez Rivas can be reached at 571-272-2589. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/GEORGE GIROUX/Primary Examiner, Art Unit 2128