DETAILED ACTION
The present Office action is in response to the amendments filed on 24 FEBRUARY 2026.
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 .
Response to Amendment
Claims 1 and 10 have been amended. No claims have been canceled or added. Claims 1-18 are pending and herein examined.
Response to Arguments
Applicant’s arguments, see Remarks, filed 24 FEBRUARY 2026, with respect to the rejection(s) of claim(s) 1 and 10 under 35 U.S.C. § 103 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of U.S. Publication No. 2022/0385907 A1 (hereinafter “Zhang”).
With regard to claim 1, previously rejected under 35 U.S.C. § 103 as being unpatentable over U.S. Publication No. 2023/0065862 A1 (hereinafter “Karabutov”) in view of U.S. Publication No. 2022/0188612 A1 (hereinafter “Park”), Applicant alleges:
“However, Karabutov and Park fail to teach or suggest the clarified feature that a received bitstream including a plurality of feature maps and a plurality of weights. According, Karabutov and Park could not reconstruct both of the plurality of feature maps and the plurality of weights from the received bitstream.” (Remarks, p. 2.)
The argument presented by Applicant suggests the clarification of the limitation a received bitstream including a plurality of feature maps and a plurality of weights, thereby reconstructing both of the plurality of feature maps and the plurality of weights is the bitstream expressly containing both the plurality of feature maps and the plurality of weights, rather than other metadata usable for determining (i.e., decoding) both the plurality of feature maps and the plurality of weights. In view of such an interpretation, Karabutov discloses receiving in a bitstream a plurality of feature maps and not a plurality of weights. See Karabutov, ¶ [0155]. However, newly cited reference Zhang expressly discloses neural network weights being transmitted from an encoder to a decoder. See Zhang, ¶ [0052]. Therefore, Karabutov and Zhang disclose every limitation, as outlined in the 35 U.S.C. § 103 rejection below.
Claim Rejections - 35 USC § 103
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) 1-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Publication No. 2023/0065862 A1 (hereinafter “Karabutov”) in view of U.S. Publication No. 2022/0385907 A1 (hereinafter “Zhang”).
Regarding claim 1, Karabutov discloses a neural processing unit (NPU) for processing a feature map (FIG. 2, decoder system 260; [0126], ll. 30-31, “a feature map is an output of a neural network layer.” Note, the feature map is the latent space depicted in FIG. 15 that is processed using a neural network), the NPU comprising:
a first circuitry provided for a feature decoder to decode a received bitstream including a plurality of feature maps([0155], “receive (obtain) a bitstream including both containers (base layer feature bitstream and enhancement layer feature bitstream) and to extract (parse) and decode only the base layer feature bitstream.” FIG. 1, base feature reconstruction subsystem 170 and enhancement feature reconstruction subsystem 140. [0215], “The base feature reconstruction subsystem 170 is configured to receive the base feature bitstream 122 and produce reconstructed base feature data 172.” [0218], “The enhancement feature reconstruction subsystem 140 is configured to receive the enhancement feature bitstream 124 and produce the reconstructed enhancement feature data 142.” [0264-0266] discloses processing circuitry); and
a second circuitry provided for at least one processing element (PE) for performing a convolution operation by using the plurality of feature maps and the plurality of weights (FIG. 1, latent space transform neural network 180, computer vision back-end network 190, and decoder neural network 150. [0115], “When programming a CNN for processing pictures or images, the input is a tensor with shape (number of images)×(image width)×(image height)×(image depth). Then, after passing through a convolutional layer, the image becomes abstracted to a feature map, with shape (number of images)×(feature map width)×(feature map height)×(feature map channels).” [0113], “Though the layers are colloquially referred to as convolutions, this is only by convention. Mathematically, it is technically a sliding dot product or cross-correlation. This has significance for the indices in the matrix, in that it affects how weight is determined at a specific index point” FIG. 14 illustrates receiving base features into a neural network with convolution. [0264-0266] discloses processing circuitry),
wherein the received bitstream includes at least one of a base layer data and at least one enhancement layer data (FIG. 4, data containers 400 and 410 representing the bitstreams. [0139], ll. 12-13, “the base layer features and enhancement layer features may be provided by the encoder in a single bitstream”),
wherein the base layer data includes a first feature map (FIG. 1, base feature data 112. FIG. 4, base feature (layer) bitstream 406. [0126], ll. 1-4, “The latent space refers to a space of features (e.g., feature maps) generated e.g. in the bottleneck layer of the trained network (e.g. a neural network) which provides data compression.” FIG. 12 depicts the neural network process of the encoder with feature data 1130 as the output),
wherein the at least one enhancement layer data includes a second feature map (FIG. 1, enhancement feature data 113. FIG. 4, enhancement feature (layer) bitstream 418. [0126], ll. 1-4, “The latent space refers to a space of features (e.g., feature maps) generated e.g. in the bottleneck layer of the trained network (e.g. a neural network) which provides data compression.” FIG. 12 depicts the neural network process of the encoder with feature data 1130 as the output)).
Karabutov fails to expressly disclose decode a received bitstream including a plurality of weights, thereby reconstructing the plurality of weights.
However, Park teaches decode a received bitstream including a plurality of weights, thereby reconstructing the plurality of weights ([0052], “a decoder may include the neural network architecture and receive the network weights from the encoder.” [0008], “receiving a compressed version of a first plurality of neural network weight values associated with a first image from a plurality of images; decompressing the first plurality of neural network weight values”).
Before the effective filing date of the claimed invention, it would have been obvious to a person having ordinary skill in the art to have signaled weights, as taught by Zhang ([0052]), in Karabutov’s invention. One would have been motivated to modify Karabutov’s invention, by incorporating Zhang’s invention, to improve coding efficiency while minimizing any degradation in video quality (Zhang: [0003]).
Regarding claim 2, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein the first feature map is related to a first artificial neural network model, and the second feature map is related to a second artificial neural network model ([0126], ll. 30-31, “a feature map is an output of a neural network layer;” FIG. 4 depicts two feature outputs; [0118] describes a plurality of neural network models capable of generating the feature map for the base layer and feature map for the enhancement layer).
Regarding claim 3, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein the first feature map is related to a kth layer of at least one artificial neural network model, and the second feature map is related to a (k-n)th layer, where k and n are integers (FIG. 15 depicts feature maps in the latent space generated by the neural network on the left side, where when expanded (see FIGS. 12 and 13) there are a plurality of layers of which each of the feature maps are “related” to. Note, the layers are in a hierarchy without any sub-layers or fractional displacement, which means the (k-n)th layer represents any previous layer and as previously stated all layers have an inherent relationship as they build upon each other in a hierarchy).
Regarding claim 4, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein the feature map is extracted based on a first area in an image, and the second feature map is extracted based on a second area in the image (FIG. 10 is an example of an input picture 202 and feature data 222 generated based on an area in an image. Note, because both feature maps are based on the image, then the “first area” and “second area” are met by the input image).
Regarding claim 5, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Zhang discloses wherein the plurality of weights in the bitstream is applied to at least one of the base layer data and the at least one enhancement layer ([0008], “processing, using a first neural network model, the first plurality of neural network weight values to yield the first image.” Paragraphs [0067-0068] and [0076-0078] describe how a neural network generates the feature map using weights. The decoding represents using the same weights on the feature maps for restoring the original images. Note, Karabutov’s disclosure describes the use of neural networks with base features and enhancement features, the disclosure of Zhang describes how said neural networks utilize the signaled weights). The same motivation of claim 1 applies to claim 5.
Regarding claim 6, Karabutov and Zhang disclose all of the limitations of claim 5, as outlined above. Additionally, Zhang discloses wherein the plurality of weights applied to at least one of the base layer data and the at least one enhancement layer data is included in the bitstream so that an additional memory for storing the plurality of weights is not needed ([0008], “receiving a compressed version of a first plurality of neural network weight values associated with a first image from a plurality of images; decompressing the first plurality of neural network weight values”). The same motivation of claim 1 applies to claim 6.
Regarding claim 7, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein at least a portion of the at least one enhancement layer data of the received bitstream is configured to be selectively processed ([0163], “The decoder systems 160 and 130 are configured to receive respectively the base feature bitstream 122 and the enhancement feature bitstream 124 and provide transformed feature data 182 and optionally a reconstructed picture 152.” [0095], “On the decoding side (e.g. cloud server), a whole latent or part thereof may be decoded selectively, as needed for human and machine vision, respectively. Thereby the bitstream is organized in a scalable manner, namely in a base layer for computer vision (object detection) and an enhancement layer for human vision”).
Regarding claim 8, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein at least a portion of the at least one enhancement layer data is configured to be selectively processed according to a preset machine analysis task ([0163], “The decoder systems 160 and 130 are configured to receive respectively the base feature bitstream 122 and the enhancement feature bitstream 124 and provide transformed feature data 182 and optionally a reconstructed picture 152.” [0095], “On the decoding side (e.g. cloud server), a whole latent or part thereof may be decoded selectively, as needed for human and machine vision, respectively. Thereby the bitstream is organized in a scalable manner, namely in a base layer for computer vision (object detection) and an enhancement layer for human vision.” Note, the machine analysis task is for human vision).
Regarding claim 9, Karabutov and Zhang disclose all of the limitations of claim 1, as outlined above. Additionally, Karabutov discloses wherein the at least one enhancement layer is included in the bitstream in ascending order according to an index of layers of the at least one enhancement layer data ([0230], “In the case of two layers, as discussed in the exemplary embodiments, one bit is used to identify the layer index (L=1 for enhancement layer).” FIG. 4 depicts L as part of the syntax in the bitstream).
Regarding claim 10, the limitations are the same as those in claim 1; however, written from the encoder perspective. Therefore, the same rationale of claim 1 applies equally as well to claim 10.
Regarding claim 11, the limitations are the same as those in claim 2; however, written from the encoder perspective. Therefore, the same rationale of claim 2 applies equally as well to claim 11.
Regarding claim 12, the limitations are the same as those in claim 3; however, written from the encoder perspective. Therefore, the same rationale of claim 3 applies equally as well to claim 12.
Regarding claim 13, the limitations are the same as those in claim 4; however, written from the encoder perspective. Therefore, the same rationale of claim 4 applies equally as well to claim 13.
Regarding claim 14, the limitations are the same as those in claim 5; however, written from the encoder perspective. Therefore, the same rationale of claim 5 applies equally as well to claim 14.
Regarding claim 15, the limitations are the same as those in claim 7; however, written from the encoder perspective. Therefore, the same rationale of claim 7 applies equally as well to claim 15.
Regarding claim 16, Karabutov and Zhang disclose all of the limitations of claim 10, as outlined above. Additionally, Karabutov discloses wherein the at least one PE is configured to process: the base layer data and a first enhancement layer data according to a first machine analysis task, or the base layer data, the first enhancement layer data and a second enhancement layer data according to a second machine analysis task ([0163], “The decoder systems 160 and 130 are configured to receive respectively the base feature bitstream 122 and the enhancement feature bitstream 124 and provide transformed feature data 182 and optionally a reconstructed picture 152.” [0095], “On the decoding side (e.g. cloud server), a whole latent or part thereof may be decoded selectively, as needed for human and machine vision, respectively. Thereby the bitstream is organized in a scalable manner, namely in a base layer for computer vision (object detection) and an enhancement layer for human vision.” Note, the machine analysis task is for human vision).
Regarding claim 17, Karabutov and Zhang disclose all of the limitations of claim 10, as outlined above. Additionally, Karabutov discloses wherein the NPU is configured to receive feedback, from the decoder, on the number of the at least one enhancement layer data included in the bitstream ([0230], “In the case of two layers, as discussed in the exemplary embodiments, one bit is used to identify the layer index (L=1 for enhancement layer) […] With these parameters {L, m}, and coded enhancement data header 414, the decoder can correctly interpret the decoded fixed point enhancement feature (layer) values and reconstruct the floating point values.” Note, the NPU is only defined by the decoder and when the decoder identifies the index value of L, then the NPU is made aware of the number of enhancement layers).
Regarding claim 18, the limitations are the same as those in claim 9; however, written from the encoder perspective. Therefore, the same rationale of claim 9 applies equally as well to claim 18.
Conclusion
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 STUART D BENNETT whose telephone number is (571)272-0677. The examiner can normally be reached Monday - Friday from 9:00 AM - 5PM 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, William Vaughn can be reached at 571-272-3922. 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.
/STUART D BENNETT/Examiner, Art Unit 2481
Prosecution Insights