DETAILED ACTION
This Office Action is in response to the amendment filed on 06/30/2026 in which Claims 1-28 are presented for examination on the merits.
Notice of Pre-AIA or AIA Status
The present application is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
1. Applicant's arguments in pages 7-9 of the REMARKS filed on 06/30/2026 with respect to the rejection under 35 USC § 103(a) and invocation of 35 U.S.C. § 112(f) have been considered. However, the corresponding arguments are moot in view of the new grounds of rejection. After careful review and in light of Applicant’s amendments, remarks, and Examiner’s newly performed search and consideration, claims 1-28 are now newly rejected under 35 U.S.C. 103(a) for the reasons specified below.
Claim Interpretation
2. The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
3. Claims 22-28 is interpreted under 35 U.S.C. 112(f) or 35 U.S.C. 112(pre-AIA ), sixth paragraph.
4. Claim limitation “means for estimating a local curvature of a loss landscape of a neural network; means for dynamically allocating parameters to be removed from the neural network based on the local curvature; and means for updating remaining weights of the neural network based on parameters to be removed.…” in claim 22 has been interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the “means” as claimed uses a generic placeholder coupled with functional language such as “dynamically allocating parameters to be removed from the neural network” without reciting sufficient structure to achieve the function. Furthermore, the generic placeholder is not preceded by a structural modifier.
5. Since these claim limitation(s) invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, claim(s) 22 has been interpreted to cover the corresponding structure described in the specification that achieves the claimed function, and equivalents thereof.
6. A review of the specification shows that the following appears to be the corresponding structure described in the specification for the 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph limitation: The blocks 702, 704, and 706 of FIG.7 illustrates “means” elements that perform dynamically allocating parameters …. updating remaining weights of the neural network to achieve corresponding function. See also the steps described in associated paragraphs 0141-0148 of the specification (Para 0145: Aspect 26: The apparatus for wireless communication of Aspects 22-25, in which the means for updating the remaining weights further comprises means for computing a single correlated weight update associated with removing all of the allocated parameters together).
If applicant wishes to provide further explanation or dispute the examiner’s interpretation of the corresponding structure, applicant must identify the corresponding structure with reference to the specification by page and line number, and to the drawing, if any, by reference characters in response to this Office action.
If applicant does not intend to have the claim limitation(s) treated under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112 , sixth paragraph, applicant may amend the claim(s) so that it/they will clearly not invoke 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, or present a sufficient showing that the claim recites/recite sufficient structure, material, or acts for performing the claimed function to preclude application of 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
For more information, see MPEP § 2173 et seq. and Supplementary Examination Guidelines for Determining Compliance With 35 U.S.C. 112 and for Treatment of Related Issues in Patent Applications, 76 FR 7162, 7167 (Feb. 9, 2011).
Claim Rejections - 35 USC § 103
7. 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 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.
8. 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.
9. 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.
10. 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.
11. Claims 1-2, 4-6, 8-9, 11-13, 15-16, 18-20, 22-23, and 25-27 are rejected under 35 U.S.C. 103 as being unpatentable over Benbaki et al. (NPL: “Fast as CHITA: Neural Network Pruning with Combinatorial Optimization”, hereinafter, Benbaki) in view of Chen et al. (WO 2022245502 A1, hereinafter, Chen).
Regarding claim 1, Durham discloses an apparatus, comprising:
at least one memory; and at least one processor coupled to the at least one memory ((§5. Conclusion: "Our single stage methods demonstrate comparable results to existing methods while achieving a significant improvement in runtime and reducing memory usage"),
the at least one processor configured to: estimate a local curvature of a loss landscape of a neural network having initial weights (§1.Introduction: "consider a quadratic approximation of the loss function based on the second-order Hessian information", and Fig.2: the claimed "estimate a local curvature of a loss landscape" corresponds to the "Hessian" "approximation");
dynamically allocate parameters to be removed from the neural network based on the local curvature ( §4. 2. Performance on gradual pruning, Table 2: "Results of gradually pruning MobilenetVl in 75% and 89% sparsity regimes, comparing CHITA to other baselines", "We also include the relative drop in accuracy to account for different methods starting from different dense weights", and §1.Introduction: "A widely used approach to mitigate inference costs is to prune or sparsify pre-trained networks by removing parameters", "In this work, we propose CHITA (Combinatorial Hessian-free Iterative Thresholding Algorithm), an efficient optimization-based framework for
network pruning at scale"); and
[update remaining weights of the neural network with a low-rank adaptation based on the parameters to be removed from the initial weights]
Benbaki does not explicitly state but Chen from the same or similar fields of endeavor teaches update remaining weights of the neural network with a low-rank adaptation based on the parameters to be removed from the initial weights (Chen, Page 6, lines 15-30: adaptation matrices allow combining the update matrices with the original weights. Adapting a large pre-trained model to specific tasks can be performed while optimizing very few parameters for the adaptation matrices wherein (Page 5, lines 33-35, Page 6, lines 1-10, Page 6, lines 22-34: a language model neural network to adapt a general model with a matrix of pretrained weights 110 for processing an input vector for adaptation matrices…obtaining neural network-based language model base model weight matrices for each of multiple neural network layers wherein raining may include the use of a loss function using standard backpropagation, calculating a gradient for every parameter and updating weights by subtracting the gradients).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to update remaining weights of the neural network with a low-rank adaptation based on the parameters to be removed from the initial weights as taught by Chen in the teachings of Benbaki for the advantage of adding first low-rank factorization matrices to corresponding base model weight matrices to form a first domain model and match fine-tuning baseline results (Chen, Page 3, lines 17-26).
Regarding claim 2, the combination of Benbaki and Chen discloses the apparatus of claim 1, in which the at least one processor is further configured to estimate the local curvature based on: weight magnitudes or activation outer products obtained from forward passes of the neural network (Benbaki §A.5. Stratified block-wise approximation: "Algorithm 6", "Obtain a feasible solution via magnitude pruning w), and
gradient outer products from backward passes of the neural network (Chen, Page 6, lines 22-34: obtaining neural network-based language model base model weight matrices for each of multiple neural network layers wherein raining may include the use of a loss function using standard backpropagation, calculating a gradient for every parameter and updating weights by subtracting the gradients).
Regarding claim 4, the combination of Benbaki and Wang discloses the apparatus of claim 1, in which the at least one processor is further configured to update the remaining weights based on assuming dependence of elements of the neural network (Benbaki, Abstract "Our approach, CHITA, extends the classical Optimal Brain Surgeon framework" and §1.lntroduction: "The recent method CBS (Combinatorial Brain Surgeon) (Yu et al., 2022) is an optimization-based approach that considers the joint effect of multiple weights).
Regarding claim 5, the combination of Benbaki and Wang discloses the apparatus of claim 1, in which the at least one processor is further configured to update the remaining weights by computing a single correlated weight update associated with removing all of the allocated parameters together (Benbaki, Para 0118, 0255: first-order optimization algorithm usually updates a parameter by using the following rule: θ2=θ1−η∇.sub.θ1. The additional optimization algorithm first multiplies V.sub.6 by an inverse matrix G.sup.−1 of an additional matrix G to generate the following update rule: θ2=θ1−ηG.sup.−1∇.sub.θ. θ1 is a parameter before updating (namely, a weight before updating). θ2 is an updated parameter (namely, an updated weight). η is a learning rate, and the learning rate can be preconfigured. ∇.sub.0 is a first-order gradient of a parameter obtained by performing first-order derivation on a loss function wherein a single iteration is far faster than that in the original KFAC algorithm…).
Regarding claim 6, the combination of Benbaki and Wang discloses the apparatus of claim 1, in which the at least one processor is further configured to iteratively estimate the local curvature and dynamically allocate the parameters to be removed (Benbaki, see §Introduction: "In this work, we propose CHITA (Combinatorial Hessianfree Iterative Thresholding Algorithm), an efficient optimization-based framework for network pruning at scale).
Regarding claim 8; Claim 8 is similar in scope to claim 1, and is therefore rejected under similar rationale.
Regarding claim 9; Claim 9 is similar in scope to claim 2, and is therefore rejected under similar rationale.
Regarding claim 11; Claim 11 is similar in scope to claim 4, and is therefore rejected under similar rationale.
Regarding claim 12; Claim 12 is similar in scope to claim 5, and is therefore rejected under similar rationale.
Regarding claim 13; Claim 13 is similar in scope to claim 6, and is therefore rejected under similar rationale.
Regarding claim 15; Claim 15 is similar in scope to claim 1, and is therefore rejected under similar rationale.
Regarding claim 16; Claim 16 is similar in scope to claim 2, and is therefore rejected under similar rationale.
Regarding claim 18; Claim 18 is similar in scope to claim 4, and is therefore rejected under similar rationale.
Regarding claim 19; Claim 19 is similar in scope to claim 5, and is therefore rejected under similar rationale.
Regarding claim 20; Claim 20 is similar in scope to claim 6, and is therefore rejected under similar rationale.
Regarding claim 22; Claim 22 is similar in scope to claim 1, and is therefore rejected under similar rationale.
Regarding claim 23; Claim 23 is similar in scope to claim 2, and is therefore rejected under similar rationale.
Regarding claim 25; Claim 25 is similar in scope to claim 4, and is therefore rejected under similar rationale.
Regarding claim 26; Claim 26 is similar in scope to claim 5, and is therefore rejected under similar rationale.
Regarding claim 27; Claim 27 is similar in scope to claim 6, and is therefore rejected under similar rationale.
12. Claims 3, 7, 10, 14, 17, 21, 24, and 28 are rejected under 35 U.S.C. 103 as being unpatentable over Benbaki et al. (NPL: “Fast as CHITA: Neural Network Pruning with Combinatorial Optimization”, hereinafter, Benbaki) in view of Chen et al. (WO 2022245502 A1, hereinafter, Chen), and further in view of Wang et al. (US 20230092453 A1, hereinafter, Wang).
Regarding claim 3, the combination of Benbaki and Chen discloses the apparatus of claim 1, however, the combination does not explicitly states but Wang from the same or similar fields of endeavor teaches in which the at least one processor is further configured to update the remaining weights based on a Kronecker-factored approximate curvature (KFAC) (Wang, Para 0254: an original (original) kronecker-factored approximate curvature (kronecker-factored approximate curvature, KFAC) algorithm is used for testing, to obtain data in a column in which the original KFAC algorithm is located in Table 4).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention wherein the at least one processor is further configured to update the remaining weights based on a Kronecker-factored approximate curvature (KFAC) as taught by Wang in the teachings of Benbaki and Chen for the advantage of providing a parameter updating method for a plurality of times through a plurality of iterations, to reduce training time of a neural network model (Wang, Para 0006-0007).
Regarding claim 7, the combination of Benbaki and Chen discloses the apparatus of claim 1, however, the combination does not explicitly states but Wang from the same or similar fields of endeavor teaches, in which the at least one processor is further configured to iteratively dynamically allocate the parameters to be removed and update the remaining weights with a low-rank adaptation (Wang, Para 0116: a training process of the neural network model is a process of continually updating a weight in the neural network model based on a plurality of iterations wherein the model is updated by using the update weight of this iteration. Last, a next iteration is performed on a model obtained after the weight is updated in this iteration, until an entire training process of the neural network model is completed).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention wherein the at least one processor is further configured to iteratively dynamically allocate the parameters to be removed and update the remaining weights with a low-rank adaptation as taught by Wang in the teachings of Benbaki and Chen for the advantage of providing a parameter updating method for a plurality of times through a plurality of iterations, to reduce training time of a neural network model (Wang, Para 0006-0007).
Regarding claim 10; Claim 10 is similar in scope to claim 3, and is therefore rejected under similar rationale.
Regarding claim 14; Claim 14 is similar in scope to claim 7, and is therefore rejected under similar rationale.
Regarding claim 17; Claim 17 is similar in scope to claim 3, and is therefore rejected under similar rationale.
Regarding claim 21; Claim 21 is similar in scope to claim 7, and is therefore rejected under similar rationale.
Regarding claim 24; Claim 24 is similar in scope to claim 3, and is therefore rejected under similar rationale.
Regarding claim 28; Claim 28 is similar in scope to claim 7, and is therefore rejected under similar rationale.
Examiner Notes
13. The Examiner notes that incorporating the combined limitations of claims 2, 5, 6, and 7 or similar claims into independent claim 1 would better clarify the subject matter/embodiment of claimed invention. Similarly, amending independent claims 8, 15, and 22 with aforesaid claim limitations from similar claims would help advance the prosecution as it would clarify the claimed invention.
Conclusion
14. The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Crabtree et al. (US 20250390352 A1) discloses a unified framework integrating an adaptive elastic funnel (AEF) with a convergent intelligence fabric (CIF) for multi-agent AI collaboration. The system provides a universal multi-modal key-value subsystem for sharing partial computations, implements hybrid placement strategies for dynamic memory management, and incorporates quantum-resistant secure enclaves.
Tran et al. (US 20250061307 A1) discloses a multi-layer artificial intelligence system includes a foundation layer comprising at least one general-purpose large language model (LLM); an expert array layer comprising a plurality of specialized reasoning models; and a meta-reasoning model configured to coordinate operations between the foundation layer and the expert array layer to generate a reasoned analysis.
Bajic (US 20250028935 A1) discloses systems which involve computer memories where a memory in accordance with this disclosure can be a multi-value memory in which each storage element of the memory can store multiple values as opposed to a standard binary storage element. The memory can include a decoder neural network and an encoder neural network to denoise the values in the memory.
15. 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 extension fee 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 MAHFUZUR RAHMAN whose telephone number is (571)270-7638. The examiner can normally be reached on Monday thru Friday.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Yin-Chen Shaw can be reached on 571-272-8878. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MAHFUZUR RAHMAN/
Primary Examiner, Art Unit 2498
Prosecution Insights