CTNF 18/740,014 CTNF 97745 Detailed Action Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Priority 02-27 AIA Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. 10-2024-0002001 , filed on January 5, 2024 with the Korean Intellectual Property Office . Information Disclosure Statement The information disclosure statement (IDS) submitted on 6/11/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. The information disclosure statement (IDS) submitted on 12/03/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. The information disclosure statement (IDS) submitted on 03/20/2025 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. The information disclosure statement (IDS) submitted on 01/21/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner. Election/Restrictions 08-25-01 AIA Applicant’s election without traverse of Group I in the reply filed on 4/16/2026 is acknowledged. 08-06 AIA Claim s 14-18 are withdrawn from further consideration pursuant to 37 CFR 1.142(b) as being drawn to a nonelected Group II , there being no allowable generic or linking claim. Election was made without traverse in the reply filed on 4/16/2026 . Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 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. 07-20-aia AIA 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. 07-21-aia AIA Claim (s) 1-10, 12, 19, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Shan et al. (US 20210089874 A1) referred to as Shan hereinafter and further in view of He et al. (Depth-wise Decomposition for Accelerating Separable Convolutions in Efficient Convolutional Neural Networks” Refer to the IDS dated 3/20/2025, refer to as He hereinafter . Regarding claim 1, Shan teaches A processor-implemented method, comprising: obtaining a first parameter of a first machine learning model; PNG media_image1.png 125 478 media_image1.png Greyscale (Shan, para. [0035]) obtaining a second parameter by transforming the first parameter based on a parameter transformation model according to a lightweighting level; (“The present invention provides an ultra-low power keyword spotting neural network circuit and a method for mapping data. A neural network model used is the depth wise separable convolutional neural network, of which a weight value and an intermediate activation value are both binarized during training, so as to obtain a lightweight neural network model with a small memory size and a small computation quantity.” Shan, para. [0007]) However, Shan does not teach generating a second machine learning model, based on the second parameter, by replacing one or more layers of the first machine learning model with a lightweight layer according to the lightweighting level; and performing an inference operation using the generated second machine learning model. He teaches generating a second machine learning model, based on the second parameter, by replacing one or more layers of the first machine learning model with a lightweight layer according to the lightweighting level; (“First we train a ”folded” ShuffleNet V2 [47]. In this model we replace depth-wise separable convolutional layers with regular convolutional layers. For example, a depth-wise convolution layer with shape C × C × k h × k w and a point-wise convolution with shape N × C × 1 × 1 will be replaced by a regular convolutional layer with shape N×C ×k h ×k w .” He, page 7, 4.3. Whole Model Decomposition) and performing an inference operation using the generated second machine learning model. (“in Figure 1 (e)), we propose to decompose an ”unseparated” convolution into a depth-wise separable convolution, while minimizing the performance degradation. Specifically, we first replace all depth-wise separable convolutions in a compact convolutional neural network (e.g., ShuffleNet V2 [47]) with standard convolutions. Then the standard convolutional neural network is trained from scratch on ImageNet. Finally, we decompose the standard convolutions into depth-wise separable convolution to obtain a new compact CNN which is of the same architecture however performs better.” He, page 2, Introduction and figure 1) Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s second machine learning. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 2, Shan does not teach wherein the first machine learning model comprises: a first convolutional layer, wherein the second machine learning model comprises: a second convolutional layer that is lighter than the first convolutional layer. He teaches wherein the first machine learning model comprises: a first convolutional layer, wherein the second machine learning model comprises: a second convolutional layer that is lighter than the first convolutional layer . (“we consider a regular convolutional layer shown in the left part of Figure 2, where the weights tensor W (n × c × k h × k w ) is applied to the input feature maps (N × c ×H ×W) which results in the output feature maps (N × n × H × W). The weights tensor can be decomposed into depth-wise convolutional weights D and pointwise convolutional weights P with shape c × c × k h × k w and n × c × 1 × 1 respectively, shown in the right part of Figure 2. In the following formulations, we do not consider the bias term for simplicity.” He, pages 3-4, 3. Approach) and fig. 2 Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s second convolutional layer. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 3, Shan does not teach wherein the second convolutional layer has a smaller number of kernel weights than the first convolutional layer. He teaches wherein the second convolutional layer has a smaller number of kernel weights than the first convolutional layer. (“Let n and c be the number of output and input channels. Let N be the number of samples. Let k h and k w be the kernel size. Let H and W be the spatial size of the feature maps.” He, page 3, 3. Approach) PNG media_image2.png 285 340 media_image2.png Greyscale He, page 4 and fig. 2 Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s second convolutional layer. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 4, Shan teaches and the second convolutional layer is a depthwise separable (DS) convolutional layer . (“A neural network model used is the depthwise separable convolutional neural network, of which a weight value and an intermediate activation value are both binarized during training, so as to obtain a lightweight neural network model with a small memory size and a small computation quantity.” Shan, para. [0007]) He teaches wherein the first convolutional layer is a standard convolutional layer (“Motivated by this, in this paper, we take the first step towards mitigating the performance degradation of depth wise separable convolution. Inspired by channel decomposition work [73] which decomposes a standard CNN to a narrower CNN without much degradation (shown in Figure 1 (e)), we propose to decompose an ”unseparated” convolution into a depth-wise separable convolution, while minimizing the performance degradation.” He, page 2) Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s standard convolutional layer. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 5, Shan teaches wherein the obtaining of the first parameter comprises: identifying the first machine learning model as a model required by an application; (“For input data of M channels, regarding the convolution in the first step, the channels are separated, and therefore, the convolution is a convolution in two-dimensional space instead of three-dimensional space, and a total scale of a depthwise separable kernel (DS kernel) is equivalent to a scale of a convolution kernel of a common convolution. A channel-separated convolution is the convolution in the first step, but a result obtained is still for the M channels.” Shan, para. [0033]) selecting a standard convolutional layer from a plurality of layers of the first machine learning model; (“The control module is mainly composed of two nesting state machines, wherein an upper-layer state machine controls interlayer skip, of which a state indicates at which layer a computation of the neural network is performed by the neural network circuit at present, and a lower-layer state machine controls specific behavior, including data loading, accumulation, bias addition, activation, output, etc., of the memory module, the data mapping module and the data processing unit array.” Shan, para. [0020]) and determining the first parameter of the selected standard convolutional layer. PNG media_image3.png 156 598 media_image3.png Greyscale Shan, para. [0035]) Regarding claim 6, Shan does not teach further comprising: determining the lightweighting level based on an available resource of an electronic device that is to perform the inference operation. He teaches further comprising: determining the lightweighting level based on an available resource of an electronic device that is to perform the inference operation. PNG media_image4.png 72 342 media_image4.png Greyscale He, page 4, 3.1. Data Processing Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s determining lightweighting level. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 7, Shan teaches identifying the first machine learning model as a model required by an application; and determining the lightweighting level based on a memory usage required for implementing the first machine learning model and an available memory of an electronic device at a time point of execution of the application . (“on the basis of a binarized depthwise separable convolutional neural network model and according to memory of a hardware circuit and characteristics of computational data, an architecture of the neural network, reduces, on the basis of ensuring the accuracy of network recognition, the memory size and the computation quantity required, so as to meet the requirement of low storage and the low computation quantity of the hardware circuit, and hereby designs a low-power keyword recognition circuit.” Shan, para. [0010]) Regarding claim 8, Shan teaches obtaining, from the lightweighting level and a plurality of first layers of the first machine learning model, information indicative of a layer to be lightened among the plurality of first layers and the second parameter to be applied to the layer to be lightened, based on the parameter transformation model. (“the neural network model used by the present invention is the depthwise separable convolutional neural network which greatly reduces the data computation quantity and the parameter memory quantity compared with a convolutional network, and of which the weight value and the intermediate activation value are both binarized during training, so as to obtain a lightweight neural network model with a small memory size and a small computation quantity;” Shan, para. [0023]) Regarding claim 9, Shan teaches wherein the first parameter is selected based on an input indicative of a layer to be lightened among a plurality of first layers of the first machine learning model. (“the neural network model used by the present invention is the depthwise separable convolutional neural network which greatly reduces the data computation quantity and the parameter memory quantity compared with a convolutional network, and of which the weight value and the intermediate activation value are both binarized during training, so as to obtain a lightweight neural network model with a small memory size and a small computation quantity;” Shan, para. [0023]) Regarding claim 10, Shan does not teach wherein the generating the second machine learning model comprises: setting the second parameter in the lightweight layer, which replaces a first layer of the first machine learning model . He teaches wherein the generating the second machine learning model comprises: setting the second parameter in the lightweight layer, which replaces a first layer of the first machine learning model . (“First we train a ”folded” ShuffleNet V2 [47]. In this model we replace depth-wise separable convolutional layers with regular convolutional layers. For example, a depth-wise convolution layer with shape C × C × k h × k w and a point-wise convolution with shape N × C × 1 × 1 will be replaced by a regular convolutional layer with shape N×C×k h ×kw.” He, page 7, 4.3. Whole Model Decomposition) Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s second machine learning. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition) Regarding claim 12, Shan does not teach wherein the second parameter is determined by providing, to the parameter transformation model, first parameters of a plurality of convolutional layers selected from the first machine learning model, in a unit of a single layer or a plurality of layers . He teaches wherein the second parameter is determined by providing, to the parameter transformation model, first parameters of a plurality of convolutional layers selected from the first machine learning model, in a unit of a single layer or a plurality of layers . (“Firstly, we want to evaluate the reconstruction error of Depth-wise Decomposition for a single layer. We conduct experiments on decomposing five different convolutional layers of ShuffleNet V2 [47]. We decompose each 3-by-3 convolutional layer with the baseline method [73], Depthwise Decomposition and Depth-wise Decomposition with error compensation. Figure 3 shows the relative error of reconstructing each of the five convolutional layers. As shown in the figure, Depth-wise Decomposition results in lower reconstruction error for all five layers.” He, page 6, 4.2. Single Layer Decomposition) and PNG media_image5.png 290 351 media_image5.png Greyscale He, page 4 Shan and He are combinable because they are from the same field of endeavor, lightweight neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan in light of He’s single layer or plurality of layers. One would have been motivated to do so because it could potentially improve the performance of current compact neural networks which favor depth-wise separable convolution a lot. (He, page 3, 2.4. Tensor Decomposition Regarding claim 19, Shan teaches A non-transitory computer-readable storage medium storing instructions that are executable by a processor to perform the method of claim 1 . (“The architecture of the neural network circuit is designed by using software and hardware in a coordinated mode, and the number of array type processing units is adapted to a size of a memory unit, to make the number of rows of each memory sub-unit and the number of the array type processing units equal to the number of channels of a convolution kernel.” Shan, para. [0016]) Regarding claim 20, refer to the explanation of claim 1 . 07-21-aia AIA Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Shan and He as mentioned above and further in view of Kang et al. (US 20240169482 A1) referred to as Kang hereinafter . Regarding claim 13, the combination of Shan and He does not teach wherein the performing of the inference operation comprises: generating a high-resolution output image from a low-resolution input image using the second machine learning model. However, Kang teaches wherein the performing of the inference operation comprises: generating a high-resolution output image from a low-resolution input image using the second machine learning model. (“The processor of the electronic device may upscale input frame images in an order of the frames. Therefore, at the time of upscaling the input frame image of the first frame (e.g., the i-th frame), the processor of the electronic device may be in a state in which upscaling has been completed for an input frame image of the second frame (e.g., the i−1-th frame). Accordingly, the processor may load an output frame image HR(i−1) (“HR” meaning “high resolution”) of the second frame (e.g., the i−1-th frame) that has been obtained by previously upscaling an input frame image LR(i−1) (“LR” meaning “low resolution”) of the second frame (e.g., the i−1-th frame) based on the first neural network and the second neural network.” Kang, para. [0065]) Shan, He, Kang are combinable because they are from the same field of endeavor, weight of neural networks. It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Shan and He in light of Kang’s generating high-resolution image. One would have been motivated to do so because it can enable images to be rendered at a low resolution (reducing the cost of rendering) and then upscaled to match an ordinary display. (Kang, para. [0003]) Allowable Subject Matter 12-151-08 AIA 07-43 12-51-08 Claim 11 is objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to PARDIS SOHRABY whose telephone number is (571)270-0809. The examiner can normally be reached Monday - Friday 9 am till 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, Jennifer Mehmood can be reached at (571) 272-2976. 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. /PARDIS SOHRABY/ Examiner, Art Unit 2664 /JENNIFER MEHMOOD/ Supervisory Patent Examiner, Art Unit 2664 Application/Control Number: 18/740,014 Page 2 Art Unit: 2664 Application/Control Number: 18/740,014 Page 3 Art Unit: 2664 Application/Control Number: 18/740,014 Page 4 Art Unit: 2664 Application/Control Number: 18/740,014 Page 5 Art Unit: 2664 Application/Control Number: 18/740,014 Page 6 Art Unit: 2664 Application/Control Number: 18/740,014 Page 7 Art Unit: 2664 Application/Control Number: 18/740,014 Page 8 Art Unit: 2664 Application/Control Number: 18/740,014 Page 9 Art Unit: 2664 Application/Control Number: 18/740,014 Page 10 Art Unit: 2664 Application/Control Number: 18/740,014 Page 11 Art Unit: 2664 Application/Control Number: 18/740,014 Page 12 Art Unit: 2664 Application/Control Number: 18/740,014 Page 13 Art Unit: 2664
Prosecution Insights