DETAILED ACTION
This office action is in response to the Application No. 18392227 filed on
12/21/2023. Claims 1-19 are presented for examination and are currently pending.
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 .
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.
3. Claims 1-19 are rejected under 35 U.S.C 101 because the claimed invention is directed towards an abstract idea without significantly more.
Step 1
Independent claim 1 is directed to a method, and falls into one of the four statutory categories.
Step 2A, Prong 1
Claim 1 recites the following abstract ideas:
finding a subnetwork of the neural network for the current task based on the binary mask (Mental process directed to finding a subnetwork of the neural network. This can be done by observing the neural network and making a judgement on the subnetwork using the binary mask).
Step 2A, Prong 2
Claim 1 recites the following additional elements:
using, in a forward pass of a neural network for learning a current task of the plurality of tasks, a plurality of weights including selected weights, the selected weights being selected in a previous task of the plurality of tasks (This limitation is directed to mere instruction to apply a judicial exception. This does not integrate the abstract idea into a practical application. See MPEP 2106.05(f));
freezing the selected weights and updating weights excluding the selected weights from the plurality of weights, in a backward pass of the neural network for learning the current task (This limitation is directed to mere instruction to apply a judicial exception. This does not integrate the abstract idea into a practical application. See MPEP 2106.05(f));
obtaining a binary mask for selecting some weights of the plurality of weights based on a weight score of each of the plurality of weights (This limitation is directed insignificant extra solution activity of data gathering. This limitation does not integrate the abstract idea into a practical application. See MPEP 2106.05(g)); and
Step 2B
Claim 1 recites the following additional elements:
using, in a forward pass of a neural network for learning a current task of the plurality of tasks, a plurality of weights including selected weights, the selected weights being selected in a previous task of the plurality of tasks (This limitation is directed to mere instruction to apply a judicial exception. This does not amount to significantly more than judicial exception. See MPEP 2106.05(f));
freezing the selected weights and updating weights excluding the selected weights from the plurality of weights, in a backward pass of the neural network for learning the current task (This limitation is directed to mere instruction to apply a judicial exception. This does not amount to significantly more than judicial exception. See MPEP 2106.05(f));
obtaining a binary mask for selecting some weights of the plurality of weights based on a weight score of each of the plurality of weights (This limitation is directed insignificant extra solution activity of data gathering and it is well understood routine and conventional. This limitation does not amount to significantly more than judicial exception. See MPEP 2106.05(d)(II), example i); and
4. Dependent claim 2 is directed to a method, and falls into one of the four statutory categories.
Claim 2 recites the following abstract ideas:
wherein the binary mask selects, as the some weights, weights whose weight
scores belong to a top-c% from among the plurality of weights, and wherein the c is a target capacity ratio (Mental process directed to selecting weight score using the binary mask. It can be done by observing the weight scores and making a judgement of the percentage to select).
Claim 2 do not recite any additional element.
5. Dependent claim 3 is directed to a method, and falls into one of the four statutory categories.
Claim 3 recites the following abstract ideas:
wherein the binary mask selects, as the some weights, the weights whose weight scores belong to the top-c% for each layer of the neural network (Mental process directed to selecting weight score using the binary mask. It can be done by observing the weight scores and making a judgement of the percentage to select).
Claim 3 do not recite any additional element.
6. Dependent claim 4 is directed to a method, and falls into one of the four statutory categories.
Claim 4 recites the following abstract ideas:
wherein the freezing the selected weights comprises freezing the selected weights and updating the weights excluding the selected weights from the plurality of weights, based on an accumulate binary mask obtained by accumulating binary masks obtained in tasks from an initial task to the previous task (Mental process directed to freezing the selected weight and updating the weight excluding the selected weight based on accumulated binary mask. This can be done by observing and making a judgement of what weight to freeze, select and update base on the accumulated binary mask).
Claim 4 do not recite any additional element.
7. Dependent claim 5 is directed to a method, and falls into one of the four statutory categories.
Claim 5 recites the following abstract ideas:
further comprising calculating a loss based on weights selected by the binary mask (Mathematical concepts directed to calculating loss based on weights selected by binary mask),
wherein the freezing the selected weights comprises freezing the selected weights and updating the weights excluding the selected weights from the plurality of weights, based on the accumulate binary mask and the loss (Mental process directed to freezing the selected weight and updating the weight excluding the selected weight based on accumulated binary mask and the loss. This can be done by observing and making a judgement of what weight to freeze, select and update base on the accumulated binary mask and the loss).
Claim 5 do not recite any additional element.
8. Dependent claim 6 is directed to a method, and falls into one of the four statutory categories.
Claim 6 recites the following abstract ideas:
further comprising: calculating a loss based on weights selected by the binary mask, and updating the weight score based on the loss (Mathematical concepts directed to calculating loss based on weights selected by binary mask and updating the weight score based on the loss).
Claim 6 do not recite any additional element.
9. Dependent claim 7 is directed to a method, and falls into one of the four statutory categories.
Claim 7 do not recite any abstract ideas.
Claim 7 recites the following additional elements:
further comprising obtaining an accumulate binary mask by accumulating binary masks obtained in tasks from an initial task to the current task among the plurality of tasks (This limitation is directed insignificant extra solution activity of data gathering. This limitation does not integrate the abstract idea into a practical application. See MPEP 2106.05(g)).
Claim 7 recites the following additional elements:
further comprising obtaining an accumulate binary mask by accumulating binary masks obtained in tasks from an initial task to the current task among the plurality of tasks (This limitation is directed insignificant extra solution activity of data gathering and it is well understood routine and conventional. This limitation does not amount to significantly more than judicial exception. See MPEP 2106.05(d)(II), example i).
10. Dependent claim 8 is directed to a method, and falls into one of the four statutory categories.
Claim 8 recites the following abstract ideas:
further comprising: converting a plurality of binary masks obtained in the plurality of tasks into a single accumulated mask (Mental process directed to converting a plurality of binary masks into a single accumulated mask. This can be done with the use of pen and paper); and
compressing the single accumulated mask into a binary map (Mental process directed to compressing the single accumulated mask using a compression algorithm. This compression algorithm can be done with a pen and paper)
Claim 8 do not recite any additional elements.
11. Dependent claim 9 is directed to a method, and falls into one of the four statutory categories.
Claim 9 recite the following abstract ideas:
wherein compressing the single accumulated mask into the binary map comprises: changing each integer of the decimal mask to an ASCII code to generate an N-bit binary mask; and compressing the N-bit binary mask using a lossless compression algorithm (Mental process directed to compressing the single accumulated mask into the binary map by changing each integer of the decimal mask to an ASCII code to generate an N-bit binary mask which can be done with a pen and paper).
Claim 9 recites the following additional elements:
wherein the single accumulated mask is a decimal mask (This limitation is directed to a particular type or source of data, which is field of use. This limitation does not integrate the abstract idea into a practical application. See MPEP 2106.05(h)), and
Claim 9 recites the following additional elements:
wherein the single accumulated mask is a decimal mask (This limitation is directed to a particular type or source of data, which is field of use. This limitation does not amount to significantly more than judicial exception. See MPEP 2106.05(h)), and
12. Independent claim 10 is directed to an apparatus, and falls into one of the four statutory categories.
With regards to claim 10, it is substantially similar to claim 1, and is rejected in the same manner and reasoning applying.
Claim 10 further recites “a memory configured to store one or more instructions; and a processor configured to, by executing one or more instructions:” these limitations are directed to a generic computer component. These limitations do not integrate the abstract idea into a practical application and do not amount to significantly more. See MPEP 2106.05(f).
13. Dependent claim 11 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 11, it is substantially similar to claim 2, and is rejected in the same manner and reasoning applying.
14. Dependent claim 12 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 12, it is substantially similar to claim 3, and is rejected in the same manner and reasoning applying.
15. Dependent claim 13 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 13, it is substantially similar to claim 4, and is rejected in the same manner and reasoning applying.
16. Dependent claim 14 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 14, it is substantially similar to claim 5, and is rejected in the same manner and reasoning applying.
17. Dependent claim 15 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 15, it is substantially similar to claim 6, and is rejected in the same manner and reasoning applying.
18. Dependent claim 16 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 16, it is substantially similar to claim 7, and is rejected in the same manner and reasoning applying.
19. Dependent claim 17 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 17, it is substantially similar to claim 8, and is rejected in the same manner and reasoning applying.
20. Dependent claim 18 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 18, it is substantially similar to claim 9, and is rejected in the same manner and reasoning applying.
21. Independent claim 19 is directed to a method, and falls into one of the four statutory categories.
With regards to claim 19, it is substantially similar to claim 1, and is rejected in the same manner and reasoning applying.
Claim 19 further recites “a computer program stored in a non-transitory computer-readable storage medium and executed by a computing device, the computer program configuring the computing device to execute:” these limitations are directed to a generic computer component. These limitations do not integrate the abstract idea into a practical application and do not amount to significantly more. See MPEP 2106.05(f)
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.
22. Claims 1-3, 10-12 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Hung et al. ("Compacting, picking and growing for unforgetting continual learning." Advances in neural information processing systems 32 (2019)) in view of Miles et al. (“Cascaded channel pruning using hierarchical self-distillation”, arXiv preprint arXiv:2008.06814, 2020)
Regarding claim 1, Hung teaches a continual learning method of learning a plurality of task in a sequential order, performed by (In this paper, we propose a simple but effective approach to continual deep learning ... we introduce an incremental learning method that is scalable to the number of sequential tasks in a continual learning process (abstract)),
the method comprising: using, in a forward pass of a neural network for learning a current task of the plurality of tasks (Task k to k+1: ... M is quantized with a threshold on ˆ M and applied to the forward pass (pg. 5, second para.); Figure 1, in the forward pass → Pick Learned Weights and Fill the Remaining Weights),
a plurality of weights (The model weights preserved for tasks 1 to k are denoted as WP 1:k. The released (redundant) weights associated with task-k are denoted as WE k, and they are extra weights that can be used for subsequent tasks, pg. 5, second para.) including selected weights (Our method is designed by combining the ideas of deep model compression via critical weights selection (Picking), pg. 2, Motivation of Our Method Design),
the selected weights being selected in a previous task of the plurality of tasks (Our approach (CPG) is accomplished by a compacting→picking(→growing) loop, which selects critical weights from old tasks, pg. 4, second to the last para.);
freezing the selected weights (Given a new task, the weights of the old-task models are fixed as well ... As the old-task weights are only picked but fixed, pg. 3, Method Overview; the old weights WP1:k (pg. 5, second para.);The Examiner notes that the picked old weights that are fixed are selected frozen weights) and
updating weights excluding the selected weights from the plurality of weights, in a backward pass of the neural network for learning the current task (The mask M and the additional weights WEk are learned together on the training data of task-(k+1) with the loss function of task-(k+1) via back-propagation, pg. 5, second para. The Examiner notes that back-propagation includes updating weights and that the selected weight which is the old weights WP1:k is not included in the update);
obtaining a binary mask for selecting some weights of the plurality of weights (Then a learnable binary weight-picking mask is trained along with previously released space for new tasks, pg. 2, Fig. 1; Hence, given a new task, we pick a set of weights (known as the critical weights) from the compact model via a learnable mask., pg. 5, second para.) based on a weight ... of each of the plurality of weights (we apply a learnable mask M to pick the old weights WP1:k, M ∈ {0,1}D with D the dimension of WP1:k. The weights picked are then represented as M ◌WP1:k, the element-wise product of the 0-1 mask M and WP1:k); and
finding a subnetwork of the neural network for the current task based on the binary mask (Compaction of task k+1: After M and WEk are learned, an initial model of task-(k+1) is obtained. Then, we fix the mask M and apply gradual pruning to compress WEk, so as to get the compact model WPk+1 ... Details of CPG continual learning is listed in Algorithm 1, pg. 5, second to the last para. The Examiner notes that the finding of the subnetwork is done by learning in the compaction of task k+1 and in algorithm 1 to find the compact model WPk+1 (subnetwork)).
Hung does not explicitly teach a computing device and obtaining a binary mask for selecting some weights of the plurality of weights based on a weight score of each of the plurality of weights;
Miles teaches computing device (All our experiments are implemented in Tensorflow with an NVIDIA 2080Ti GPU, pg. 7, section 4.1 Implementation details)
obtaining a binary mask for selecting some weights of the plurality of weights (The pruning objective can be described through the use of a binary mask M ∈ {0,1}|W| that is applied to the weights ... The objective of pruning is then to learn a small subset of weights that can achieve comparable performance to the original model, pg. 4, Section 3.1, Formulation) based on a weight score of each of the plurality of weights (The binary mask M disables the least "important" weights. To identify these weights we introduce an importance score γ ∈ IR|W|, pg.4, section 3.2);
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung to incorporate the teachings of Miles for the benefit of contributing most significantly to the overall computational cost (pg. 4, Section 3.1)
Regarding claim 2, Hung and Miles teaches the method of claim 1, Miles teaches wherein the binary mask selects, as the some weights, weights whose weight scores belong to a top-c% from among the plurality of weights, and wherein the c is a target capacity ratio (We model the importance score γ as a differentiable weight for which we can compute the binary mask M. A pruning threshold is then defined as the smallest top-p% of the importance scores across all the convolutional layers, pg.4, Section 3.2).
The same motivation to combine independent claim 1 applies here.
Regarding claim 3, Hung and Miles teaches the method of claim 2, Miles teaches wherein the binary mask selects, as the some weights, the weights whose weight scores belong to the top-c% for each layer of the neural network (We model the importance score γ as a differentiable weight for which we can compute the binary mask M. A pruning threshold is then defined as the smallest top-p% of the importance scores across all the convolutional layers, pg.4, Section 3.2).
The same motivation to combine independent claim 1 applies here.
Regarding claim 10, claim 10 is similar to claim 1. It is rejected in the same manner and reasoning applying. Further, Miles teaches a continual learning apparatus comprising: a memory configured to store one or more instructions; and a processor configured to, by executing one or more instructions (All our experiments are implemented in Tensorflow with an NVIDIA 2080Ti GPU, pg. 7, section 4.1 Implementation details. The Examiner notes that the GPU includes a memory the store instructions):
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung to incorporate the teachings of Miles for the benefit of contributing most significantly to the overall computational cost (pg. 4, Section 3.1)
Regarding claim 11, claim 11 is similar to claim 2. It is rejected in the same manner and reasoning applying.
Regarding claim 12, claim 12 is similar to claim 3. It is rejected in the same manner and reasoning applying.
Regarding claim 19, claim 19 is similar to claim 1. It is rejected in the same manner and reasoning applying. Further, Miles teaches a computer program stored in a non-transitory computer-readable storage medium and executed by a computing device, the computer program configuring the computing device to execute: (All our experiments are implemented in Tensorflow with an NVIDIA 2080Ti GPU, pg. 7, section 4.1 Implementation details. The Examiner notes that the GPU includes a memory the store instructions):
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung to incorporate the teachings of Miles for the benefit of contributing most significantly to the overall computational cost (pg. 4, Section 3.1)
23. Claims 4-7 and 13-16 are rejected under 35 U.S.C. 103 as being unpatentable over Hung et al. ("Compacting, picking and growing for unforgetting continual learning." Advances in neural information processing systems 32 (2019)) in view of Miles et al. (“Cascaded channel pruning using hierarchical self-distillation”, arXiv preprint arXiv:2008.06814, 2020) and further in view of Wan et al. ("Regularization of neural networks using dropconnect." International conference on machine learning. PMLR, 2013).
Regarding claim 4, Hung and Miles teaches the method of claim 1, Hung teaches wherein the freezing the selected weights comprises freezing the selected weights (Given a new task, the weights of the old-task models are fixed as well ... As the old-task weights are only picked but fixed, pg. 3, Method Overview; the old weights WP1:k (pg. 5, second para.);The Examiner notes that the picked old weights that are fixed are selected frozen weights) and updating the weights excluding the selected weights from the plurality of weights (The mask M and the additional weights WEk are learned together on the training data of task-(k+1) with the loss function of task-(k+1) via back-propagation, pg. 5, second para. The Examiner notes that back-propagation includes updating weights and that the selected weight which is the old weights WP1:k is not included in the update),
Hung and Miles does not explicitly teach updating the weights ... from the plurality of weights, based on an accumulate binary mask obtained by accumulating binary masks obtained in tasks from an initial task to the previous task.
Wan teaches updating the weights ... from the plurality of weights, based on an accumulate binary mask obtained by accumulating binary masks obtained in tasks from an initial task to the previous task (The overall model f(x;θ,M) therefore maps input data x to an output o through a sequence of operations given the parameters θ = {Wg,W,Ws} and randomly drawn mask M. The correct value of o is obtained by summing out over all possible masks M:
PNG
media_image1.png
52
394
media_image1.png
Greyscale
... where the output is a mixture of 2|M| different networks, each with weight p(M), pg. 3, right col., to pg. 4, left col., section 3. Model Description, 4. Cross Entropy Loss).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung and Miles to incorporate the teachings of Wan for the benefit of improving the network’s generalization ability (Wan, Section 2.1)
Regarding claim 5, Hung, Miles and Wan teaches the method of claim 4, Hung teaches wherein the freezing the selected weights comprises freezing the selected weights (Given a new task, the weights of the old-task models are fixed as well ... As the old-task weights are only picked but fixed, pg. 3, Method Overview; the old weights WP1:k (pg. 5, second para.);The Examiner notes that the picked old weights that are fixed are selected frozen weights) and
updating the weights excluding the selected weights from the plurality of weights(The mask M and the additional weights WEk are learned together on the training data of task-(k+1) with the loss function of task-(k+1) via back-propagation, pg. 5, second para. The Examiner notes that back-propagation includes updating weights and that the selected weight which is the old weights WP1:k is not included in the update),
Wan teaches further comprising calculating a loss based on weights selected by the binary mask (Cross Entropy Loss:
PNG
media_image2.png
26
190
media_image2.png
Greyscale
takes probabilities o and the ground truth labels y as input. The overall model f(x;θ,M) therefore maps input data x to an output o through a sequence of operations given the parameters θ = {Wg,W,Ws} and randomly drawn mask M, pg. 3, right col., to pg. 4, left col., section 3. Model Description, 4. Cross Entropy Loss),
updating the weights ... from the plurality of weights, based on an accumulate binary mask obtained by accumulating binary masks and the loss (Cross Entropy Loss:
PNG
media_image2.png
26
190
media_image2.png
Greyscale
takes probabilities o and the ground truth labels y as input. The overall model f(x;θ,M) therefore maps input data x to an output o through a sequence of operations given the parameters θ = {Wg,W,Ws} and randomly drawn mask M. The correct value of o is obtained by summing out over all possible masks M:
PNG
media_image1.png
52
394
media_image1.png
Greyscale
... where the output is a mixture of 2|M| different networks, each with weight p(M), pg. 3, right col., to pg. 4, left col., section 3. Model Description, 4. Cross Entropy Loss).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung and Miles to incorporate the teachings of Wan for the benefit of improving the network’s generalization ability (Wan, Section 2.1)
Regarding claim 6, Hung and Miles teaches the method of claim 1, Hung and Miles does not explicitly teach the limitation of 6.
Wan teaches further comprising: calculating a loss based on weights selected by the binary mask (Cross Entropy Loss:
PNG
media_image2.png
26
190
media_image2.png
Greyscale
takes probabilities o and the ground truth labels y as input. The overall model f(x;θ,M) therefore maps input data x to an output o through a sequence of operations given the parameters θ = {Wg,W,Ws} and randomly drawn mask M, pg. 3, right col., to pg. 4, left col., section 3. Model Description, 4. Cross Entropy Loss), and
updating the weight score based on the loss (
PNG
media_image3.png
118
382
media_image3.png
Greyscale
pg. 3, Algorithm 1).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung and Miles to incorporate the teachings of Wan for the benefit of improving the network’s generalization ability (Wan, Section 2.1)
Regarding claim 7, Hung and Miles teaches the method of claim 1, Hung and Miles does not explicitly teach the limitation of 7.
Wan teaches further comprising obtaining an accumulate binary mask by accumulating binary masks obtained in tasks from an initial task to the current task among the plurality of tasks (The correct value of o is obtained by summing out over all possible masks M:
PNG
media_image1.png
52
394
media_image1.png
Greyscale
... where the output is a mixture of 2|M| different networks, each with weight p(M), pg. 3, right col., to pg. 4, left col., section 3. Model Description, 4. Cross Entropy Loss.) .
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung and Miles to incorporate the teachings of Wan for the benefit of improving the network’s generalization ability (Wan, Section 2.1)
Regarding claim 13, claim 13 is similar to claim 4. It is rejected in the same manner and reasoning applying.
Regarding claim 14, claim 14 is similar to claim 5. It is rejected in the same manner and reasoning applying.
Regarding claim 15, claim 15 is similar to claim 6. It is rejected in the same manner and reasoning applying.
Regarding claim 16, claim 16 is similar to claim 7. It is rejected in the same manner and reasoning applying.
24. Claims 8, 9, 17 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Hung et al. ("Compacting, picking and growing for unforgetting continual learning." Advances in neural information processing systems 32 (2019)) in view of Miles et al. (“Cascaded channel pruning using hierarchical self-distillation”, arXiv preprint arXiv:2008.06814, 2020) and further in view of Gallo et al. (US20220153262)
Regarding claim 8, Hung and Miles teaches the method of claim 1, Hung and Miles does not explicitly teach the limitations of claim 8.
Gallo teaches further comprising: converting a plurality of binary masks obtained in the plurality of tasks into a single accumulated mask (a neural network 206 processes a source image 202 and a reference image 204 in connection with a scaling factor 208 to determine a binary mask 210 [0066]; In at least one embodiment, pixels of a binary mask 210 correspond to objects of reference image 204 [0076]); and
compressing the single accumulated mask into a binary map (In at least one embodiment, scaled source image features 310 and reference image features 308 are input to a segmentation network 316, which outputs to a segmentation refine network 318 (or also referred to herein as segmentation refine or SegRefine network) to generate a binary segmentation map 320 [0078]; In at least one embodiment, compression logic can be lossless compression logic that makes use of one or more of multiple compression algorithms [0364]).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung, Miles and Wan to incorporate the teachings of Gallo for the benefit of training speed for deep neural networks (Gallo [0336])
Regarding claim 9, Hung, Miles and Gallo teaches the method of claim 8, Gallo teaches wherein the single accumulated mask is a decimal mask, and wherein compressing the single accumulated mask into the binary map comprises: changing each integer of the decimal mask to an ASCII code to generate an N-bit binary mask (In at least one embodiment, if a ratio of a scale of an object of reference image 502B to a scale of a corresponding object (e.g., same object) of scaled source image 502A is greater than 1, neural network 504 assigns a value of 1 to corresponding pixels of said object in binary segmentation map 508 ... In at least one embodiment, a pixel with a value 1 corresponds to a red-green-blue (RGB) color model color white (e.g., RGB decimal code (255, 255, 255)) and a pixel with a value 0 corresponds to a RGB color model color black (e.g., RGB decimal code (0,0,0)) [0098]); and
compressing the N-bit binary mask using a lossless compression algorithm (In at least one embodiment, compression logic can be lossless compression logic that makes use of one or more of multiple compression algorithms [0364]).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Hung and Miles to incorporate the teachings of Gallo for the benefit of training speed for deep neural networks (Gallo [0336])
Regarding claim 17, claim 17 is similar to claim 8. It is rejected in the same manner and reasoning applying.
Regarding claim 18, claim 18 is similar to claim 9. It is rejected in the same manner and reasoning applying.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MORIAM MOSUNMOLA GODO whose telephone number is (571)272-8670. The examiner can normally be reached Monday-Friday 8:00am-5:00pm 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, Michelle T. Bechtold can be reached on (571) 431-0762. 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.
/M.G./Examiner, Art Unit 2148
/MICHELLE T BECHTOLD/Supervisory Patent Examiner, Art Unit 2148
Prosecution Insights