Prosecution Insights
Last updated: April 19, 2026
Application No. 18/365,047

NORMALIZATION SCHEME FOR SELF-ATTENTION NEURAL NETWORKS

Non-Final OA §101§102
Filed
Aug 03, 2023
Examiner
SHANMUGASUNDARAM, KANNAN
Art Unit
2168
Tech Center
2100 — Computer Architecture & Software
Assignee
Huawei Technologies Co., Ltd.
OA Round
1 (Non-Final)
72%
Grant Probability
Favorable
1-2
OA Rounds
3y 9m
To Grant
99%
With Interview

Examiner Intelligence

Grants 72% — above average
72%
Career Allow Rate
416 granted / 579 resolved
+16.8% vs TC avg
Strong +37% interview lift
Without
With
+37.0%
Interview Lift
resolved cases with interview
Typical timeline
3y 9m
Avg Prosecution
24 currently pending
Career history
603
Total Applications
across all art units

Statute-Specific Performance

§101
12.2%
-27.8% vs TC avg
§103
48.8%
+8.8% vs TC avg
§102
26.0%
-14.0% vs TC avg
§112
6.3%
-33.7% vs TC avg
Black line = Tech Center average estimate • Based on career data from 579 resolved cases

Office Action

§101 §102
DETAILED ACTION Claims 1-16 are pending in the Instant Application. Claims 1-16 are rejected (Non-Final Rejection). 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 The Instant Application, filed 08/03/2023, is a continuation of PCT/EP2021/052679, filed 02/04/2021, and thus has an effective filing date of 02/04/2021 for what is described therein. Information Disclosure Statement The information disclosure statement (IDS) submitted on 10 July 2024 and 07 January 2025 were considered by the examiner. Claim Objections Claim 16 is objected to because of the following informalities: claim 16 refers to claim 13, but states “any of claim 13.” Since there is only one claim 13, it appears to be a typographical error. Appropriate correction is required. 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 non-statutory subject matter. Claims 1-12 recite “a data processing system” that can be broadly interpreted as non-statutory, “software per se”. Claims 1-12 do not expressly recite hardware. The specification at [0028] states, “According to a third aspect there is provided a computer program which, when executed by a computer, causes the computer to perform the method described above.” Thus, the system can be “a computer program,” allowing it to be interpreted as “software per se”. ”Software per se” is considered non-statutory subject matter. Please include hardware such as a “processor” or a “memory” to overcome this rejection. Claim Rejections - 35 USC § 102 The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action: A person shall be entitled to a patent unless – (a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention. Claims 1-16 are rejected under 35 U.S.C. 102(a)(1) as being unpatentable by Veličković et al. (“Velickovic”), “Graph Attention Networks,” 2018. As per claim 1, Velickovic discloses a data processing device for performing an attention-based operation on a graph neural network, the device being configured to receive one or more input graphs each having a plurality of nodes and to, for at least one of the input graphs ([Page 6, 3.1 Datasets] wherein Inducive learning with 20 graphs is described, with an average of 2372 nodes per graph): form an input node representation for each node in the respective input graph ([Page 3, 2.1 Graph attentional layer] wherein an input node representation, where j is a node and is represented by its feature\ PNG media_image1.png 24 182 media_image1.png Greyscale ), wherein a respective norm is defined for each input node representation ([Page 3, 2.1 Graph attentional layer] wherein features, h, are defined for each input node as a vector, requires a respective norm); multiply each of the input node representations with each of the set of attention parameters to form a score function of the respective input graph ([Page 3, 2.1 Graph attentional layer] wherein the input node representations (input features) are multiple by the weight matrix to form a score function representative of the importance of node j’s features to node i); normalize the score function based on a maximum of the norms of the input node representations to form a normalised score function ([Page 3, 2.1 Graph attentional layer] wherein we normalize the score function using softmax for input node j); and form a weighted node representation by weighting each node in the respective input graph using a respective element of the normalised score function ([Page 3, 2.1 Graph attentional layer] wherein so the coefficients form a weighted node representation that can be compared) . As per claim 2, Velickovic discloses the data processing device of claim 1, wherein the score function is normalized such that the elements of the normalized score function sum to 1 ([Page 3, 2.1 Graph attentional layer] wherein softmax is described which ensures the resulting values are in the range of (0,1) and add up to 1). As per claim 3, Velickovic discloses the data processing device of claim 1, wherein an attention mechanism of the graph neural network is Lipschitz continuous ([Page 3, 2.1 Graph attentional layer] wherein softmax is described which is Lipschitz continuous around ½). As per claim 4, Velickovic discloses the data processing device of claim 1, wherein a softmax function is applied to the normalized score function ([Page 3, 2.1 Graph attentional layer] wherein softmax is described to normalize the score function). As per claim 5, Velickovic discloses the data processing device of claim 1, wherein a softmax function is applied to the score function of each node of the graph and the neighbouring nodes of each respective node, such that a set of score function values of each neighborhood sum to 1 ([Page 5, 2.2 Comparisons to related work] wherein Softmax can be performed on an entire neighborhood of a node, wherein the function values would then sum to 1). As per claim 6, Velickovic discloses the data processing device of claim 1, wherein the input node representation gives contextual information about the respective node ([Page 3, 2.1 Graph attentional layer] wherein the feature vectors encode contextual information). As per claim 7, Velickovic discloses the data processing device of claim 6, wherein the contextual information is in the form of a tensor ([Page 3, 2.1 Graph attentional layer] wherein the feature vectors are a type of tensor). As per claim 8, Velickovic discloses the data processing device of claim 1, wherein for each node, the respective element of the normalised score function is combined with the input representation of the respective node using a dot-product to form the weighted node representation of the node based on the weighted representation of its neighboring nodes ([Pages 3-4. 2.1 Graph attentional layer] wherein aij is the normalized score function, which is combined with the input representation hj using dot product, which is multiplication in the prior art as shown on page 4, equation 3.) As per claim 9, Velickovic discloses the data processing device of claim 1, wherein the graph neural network is a graph attention network or a graph transformer ([Page 1, Abstract] wherein a graph attention network is described). As per claim 10, Velickovic discloses the data processing device of claim 1, wherein an attention mechanism of the graph neural network comprises a multi-head attention mechanism ([Pages 4. 2.1 Graph attentional layer] wherein multi-head attention is applied). As per claim 11, Velickovic discloses the data processing device of claim 10, wherein the score function is normalized for every attention head in the multi-head attention mechanism ([Pages 3-4. 2.1 Graph attentional layer] wherein the normalized score function and performed on each attention head). As per claim 12, Velickovic disclose she data processing device of claim 1, wherein the system is configured to learn the attention parameters ([Pages 3, 1. Introduction] where in the is configured to learn the attention parameters). As per claim 13, claim 13 is the method performed by the system of claim 1 and is rejected for the same rationale and reasoning. As per claim 14, claim 14 is the method performed by the system of claim 2 and is rejected for the same rationale and reasoning. As per claim 15, claim 15 is the method performed by the system of claim 3 and is rejected for the same rationale and reasoning. As per claim 16, claim 16 is the non-transitory computer readable medium storing a computer program which, when executed by a computer, causes the computer to perform the method of any of claim 13. Thus, claim 16 is rejected for the same rationale and reasoning as claim 1 and 13. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to KANNAN SHANMUGASUNDARAM whose telephone number is (571)270-7763. The examiner can normally be reached M-F 9:00 AM -6:00 PM. 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, Charles Rones can be reached at (571) 272-4085. 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. /KANNAN SHANMUGASUNDARAM/Primary Examiner, Art Unit 2168
Read full office action

Prosecution Timeline

Aug 03, 2023
Application Filed
Mar 23, 2026
Non-Final Rejection — §101, §102 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12596756
PERSONALIZED RELATED QUERIES FOR SEARCH SEGMENTS
2y 5m to grant Granted Apr 07, 2026
Patent 12596729
Value-Directed Parsing for Data Extraction
2y 5m to grant Granted Apr 07, 2026
Patent 12585703
METHOD AND APPARATUS FOR PROCESSING GRAPH DATA, DEVICE, STORAGE MEDIUM, AND PROGRAM PRODUCT
2y 5m to grant Granted Mar 24, 2026
Patent 12579180
EVALUATION SUPPORT PROGRAM, EVALUATION SUPPORT METHOD, AND EVALUATION SUPPORT DEVICE
2y 5m to grant Granted Mar 17, 2026
Patent 12572520
MACHINE-LEARNING-AUTOMATED RECOGNITION AND LABELLING OF COLUMNAR DATA
2y 5m to grant Granted Mar 10, 2026
Study what changed to get past this examiner. Based on 5 most recent grants.

AI Strategy Recommendation

Get an AI-powered prosecution strategy using examiner precedents, rejection analysis, and claim mapping.
Powered by AI — typically takes 5-10 seconds

Prosecution Projections

1-2
Expected OA Rounds
72%
Grant Probability
99%
With Interview (+37.0%)
3y 9m
Median Time to Grant
Low
PTA Risk
Based on 579 resolved cases by this examiner. Grant probability derived from career allow rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month