DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 19-38 are pending.
Claims 1-18 are canceled.
Response to Arguments (Election/Restrictions)
Applicant's election with traverse of Group I (claims 19-27 and 37-38) in the reply filed on 6/2/2026 is acknowledged. The traversal is on the ground(s) that
the applicant asserts the Office failed to prove the sub-combinations are distinct under MPEP § 806.05(d) because it omitted a comparative analysis demonstrating non-overlapping scopes;
the Office did not establish a separate utility for either sub-combination outside the disclosed combination; and
the applicant argues the Office failed to show a serious search burden under MPEP § 808.02. Because both groups share primary classifications and feature closely related subgroups, a separate field of search is unnecessary.
This is not found persuasive. While the applicant traverses the restriction requirement, the requirement remains proper because the sub-combinations are distinct and a combined examination would impose a serious search burden. Regarding the applicant's arguments on scope and utility, MPEP § 806.05(d) states that sub-combinations may be restricted if they "do not overlap in scope and are not obvious variants, and if it is shown that at least one sub-combination is separately usable." Group I and Group II possess entirely non-overlapping scopes because they define mutually exclusive functional roles: Group I is strictly limited to an apparatus generating and transmitting data, whereas Group II is strictly limited to an apparatus receiving and further processing that data. Furthermore, the requirement of separate utility is met because the MPEP requires only a showing that a sub-combination has utility "other than in the disclosed combination." The edge device of Group I can independently operate to transmit data to generic, non-neural-network storage systems, satisfying this standard. Finally, regarding the search burden, MPEP § 808.02 states that a serious burden can be demonstrated by establishing "A separate status in the art when they are classifiable together" or "A different field of search." Even though the groups share a high-level classification, the distinct operational domains of localized sensor data generation (H04L67/12) versus centralized cloud load-balancing (H04L67/10) represent different fields of search that necessitate fundamentally divergent prior art queries, thereby justifying the restriction.
The requirement is still deemed proper and is therefore made FINAL.
Claim Rejections - 35 USC § 103
The following is a quotation of pre-AIA 35 U.S.C. 103(a) which forms the basis for all obviousness rejections set forth in this Office action:
(a) A patent may not be obtained though the invention is not identically disclosed or described as set forth in section 102 of this title, if the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains. Patentability shall not be negatived by the manner in which the invention was made.
Claim(s) 19-27 and 37-38 is/are rejected under 35 U.S.C. 103 as being unpatentable over Yang et al (US20170076195A1) in view of Wu et al (US20210406674A1).
Regarding claims 19 and 37, Yang teaches a data processing system comprising a first data processing device to execute part of a neural network by performing operations comprising:
(Yang, "distributed neural network framework 200 may include a camera 201 having a camera module 211 and a lower level layer (LLL) module 212", [0034]; a data processing system comprising a first data processing device (camera 201) configured to execute a part of a neural network (the lower level layer module))
generating a data stream via a data stream source associated with the first data processing device;
(Yang, "camera module 401 may attain an image or video of a scene and camera module 401 may generate image data 411.", [0052]; generating a data stream (image data 411) via a data stream source (camera module 401) associated with the first data processing device (camera 400))
executing, by at least one first neural network processor of the first data processing device, a first sub-network of the neural network to process the data stream and to provide a processed data stream; and
(Yang, "image data 411 may be provided to hardware accelerator 402, which may generate sub-sampled feature maps (SSFMs) 412.", [0054]; Wu, "vision, radar, and LiDAR processing chipsets can include feature extraction processing, such as special purpose ASICs and the like.", [0030]; Yang teaches executing, via a hardware accelerator acting as the first neural network processor, a first sub-network to process the data stream and provide a processed data stream (sub-sampled feature maps). Wu teaches that processing chipsets can include special purpose ASICs for feature extraction processing)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to incorporate the teachings of Wu into the system or method of Yang in order to implement the first neural network processor as a highly specialized ASIC to further optimize edge processing performance. The combination of Yang and Wu also teaches other enhanced capabilities.
The combination of Yang and Wu further teaches:
transmitting, by the first data processing device, the processed data stream to a second data processing device via a data communication channel,
(Yang, "sub-sampled feature maps 412 may be provided to transmitter 403, which may transmit sub-sampled feature maps 413 to another device (e.g., a gateway or cloud computing device or the like) using any suitable communications channel", [0058]; transmitting, by the first data processing device, the processed data stream (sub-sampled feature maps 413) to a second data processing device via a data communication channel)
wherein a second sub-network of the neural network is executed by at least one second neural network processor of the second data processing device to further process the processed data stream and to provide a further processed data stream.
(Yang, "cloud computing resource 203 may, via fully connected portion module 232 implement a fully connected portion (e.g., fully connected portion 105 or the like) of distributed neural network framework 200 to generate object label 253.", [0039]; Wu, "The fusion layers use the aggregated feature map to make decisions for bounding box regression, classification and semantic/instance segmentation.", [0033]; Yang teaches a second sub-network (fully connected portion module) executed by a second data processing device (cloud computing resource) to further process the data stream into a further processed data stream (object label). Wu teaches utilizing fusion layers at a central processor to make sophisticated decisions like semantic segmentation and bounding box regression based on the feature maps. Incorporating Wu into Yang would enhance the second sub-network to output complex fusion layer data streams like bounding box regressions and segmentations)
Regarding claim 20, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 19, wherein the data stream comprises a sequence of image frames
(Yang, "As used herein, image data may include any suitable still image data, video frame data, or the like.", [0021]; "providing a raw video stream from a 2 megapixel (MP) camera operating at 25 frames per second (FPS)", [0062]; the data stream may include a raw video stream containing video frame data, which inherently comprises a sequence of image frames)
Regarding claim 21, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 19, wherein the data stream is received by the first sub-network as successive static images and the processed data stream is rendered as successive feature maps associated with the successive static images.
(Yang, "Process 600 may be repeated any number of times either in series or in parallel for any number input images (e.g., still images or video frames)", [0081]; "lower level convolutional layers and/or sub-sampling layers may be implemented based on the image data to generate one or more convolutional neural network feature maps", [0076]; repeating the processing for any number of successive input video frames (static images) to compute and render associated successive convolutional neural network feature maps for those frames)
Regarding claims 22 and 38, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 19, the operations comprising:
computing, by the first data processing device, a stream of feature map frames for a feature map at an output of the first sub-network; and
performing, by the first data processing device, temporal compression by transmitting information about a feature map value of a feature map element of a particular feature map frame of the stream of feature map frames only in response to determining that a difference between the feature map value and an associated feature map value of the feature map element in a previous feature map frame of the stream of feature map frames meets or exceeds a threshold.
(Yang, "hardware accelerator 402 may implement compression of generated sub-sampled feature maps 412", [0057]; "providing a raw video stream from a 2 megapixel (MP) camera operating at 25 frames per second (FPS) may require a bandwidth of about >8 megabits per second (Mbps) using H.264 video coding", [0062]; computing a stream of feature maps and applying compression. computing a stream of feature maps and applying compression to reduce network bandwidth. While Yang does not explicitly detail the exact thresholding steps for the feature maps, it explicitly references H.264 video coding. H.264 inherently utilizes delta compression (inter-frame prediction), which fundamentally relies on comparing a current frame to a previous frame and transmitting only the temporal differences (residuals) that meet or exceed a specific quantization threshold. It would have been obvious to a person of ordinary skill in the art to implement the well-known H.264 delta compression technique referenced in Yang to compress the continuous stream of feature map frames. Applying this known temporal compression method to the feature maps would yield the predictable result of optimally minimizing transmission bandwidth and reducing network payload)
Regarding claim 23, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 22, wherein the previous feature map frame is an immediately preceding feature map frame relative to the particular feature map frame.
(Yang, see comments on claim 22; In standard H.264 delta compression, the most fundamental and universally applied form of inter-frame prediction (e.g., generating standard P-frames) calculates the temporal difference by comparing the current frame directly to the immediately preceding reference frame. It would be obvious to a person of ordinary skill in the art implementing the H.264 delta compression logic for Yang's feature map stream to use the immediately preceding feature map frame as the reference frame to efficiently calculate the temporal delta)
Regarding claim 24, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 22, wherein the previous feature map frame is a most recent feature map frame for which information for the feature map element was transmitted by the first data processing device.
(Yang, see comments on claim 22; In H.264 delta compression algorithms, to prevent encoder-decoder drift and error accumulation over time, the temporal difference must be calculated against a reconstructed version of the most recently transmitted reference frame, rather than simply an un-transmitted internal frame. It would be obvious to a person of ordinary skill in the art to employ this foundational rule of H.264 delta compression to ensure the fidelity of the transmitted feature map stream in Yang)
Regarding claim 25, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 22, wherein the information transmitted by the first data processing device is indicative of the difference between the feature map value and the associated feature map value of the feature map element in the previous feature map frame.
(Yang, see comments on claim 22; The core mechanism of H.264 delta compression is transmitting the "residual"—which is exactly the calculated difference (the delta) between the current frame values and the previous reference frame values—rather than transmitting the absolute values of the current frame. It would be obvious to a person of ordinary skill in the art applying H.264 delta compression to Yang's feature maps to transmit information indicative of this difference to achieve the highly efficient bandwidth reduction taught by Yang)
Regarding claim 26, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 19,
wherein the data stream comprises a first stream of frames and the first data processing device further generates a second stream of frames via the data stream source,
wherein the neural network is a first neural network, wherein still scene information is extracted by processing the first stream of frames via the first sub-network and the second sub-network of the first neural network, and
wherein spatio-temporal modeling information is extracted by processing the second stream of frames via a first sub-network of a second neural network and a second sub-network of the second neural network, the first sub- network of the second neural network being processed by the first data processing device and the second sub-network of the second neural network being processed by the second data processing device.
(Yang, "distributed neural network may include any suitable neural network such as a convolutional neural network (CNN), a deep neural network (DNN), recurrent convolutional neural network (RCNN), or the like.", [0020]; Wu, "Each sensor module can provide multiple CNNs 350 for extracting different feature sets for a variety of applications.", [0033]; Yang teaches using different network types like CNNs (for spatial/still scenes) and RCNNs (for spatio-temporal sequences). Wu teaches generating multiple streams of frames and using multiple neural sub-networks to extract different feature sets in parallel at the edge device. Together Yang and Wu teach processing parallel streams using parallel neural networks. Incorporating Wu into Yang would allow simultaneously processing multiple frame streams using respective dedicated sub-networks (e.g., CNN for still scene, RCNN for spatio-temporal) to achieve comprehensive feature extraction)
Regarding claim 27, the combination of Yang and Wu teaches its/their respective base claim(s).
The combination further teaches the data processing system of claim 19, wherein the first data processing device is positioned at a location to be monitored, and the second data processing device is positioned remote from the location."
(Yang, "For example, surveillance cameras are common on street corners, at road intersections, in parking lots, at stores, surrounding private property", [0002]; "If a computational resource remote from the camera (e.g., a cloud resource or the like) were used", [0003]; the first data processing device (surveillance camera) is positioned at a physical location to be monitored, and the second data processing device (cloud resource) positioned remote from the camera location)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JIANXUN YANG whose telephone number is (571)272-9874. The examiner can normally be reached on MON-FRI: 8AM-5PM Pacific Time.
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, Amandeep Saini can be reached on (571)272-3382. 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.
/JIANXUN YANG/
Primary Examiner, Art Unit 2662 7/25/2026
Prosecution Insights