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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 12/05/2025 has been entered.
Detailed Action
The following action is in response to the communication(s) received on 12/05/2025 after a final rejection.
Claim 1 has been amended.
Claims 2, 16, and 17 have been canceled.
Claims 1, 3-15, and 18 are now pending.
Claim 1 is an independent claim.
Response to Arguments
Applicant’s arguments filed 12/05/2025 have been fully considered, but are not fully persuasive.
Applicant asserts that the claims are in condition for allowance in view of the amendment to the claim “removing the first input datum from the first set of training data…” Examiner respectfully disagrees, as Birodkar [p.3 1st col 2nd ¶] remains teaching this limitation; discarding clusters that are the farthest from the cluster center corresponds to removing the first input data that exceed the threshold.
Thus, the claims remain unpatentable.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1, 3-15, and 18 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim 1 recites a method, thus a process, one of the four statutory categories of patentable subject matter (Step 1). However, Claim 1 further recites:
at least part of the training data forms a temporal sequence and is combined in a first set of training data, and the encoder maps input data to prototype feature vectors of a set of prototype feature vectors...; and depending on the assigned feature vectors, a defined set of prototype feature vectors is determined and assigned to the first input datum; c) creating an aggregate vector for the first input datum; ... and creating a second aggregate vector for the second input datum…; e) comparing at least the first and second aggregated vectors and determining a measure of similarity for the aggregated vectors; and f) removing the first input datum from the first set of training data when the determined measure of similarity exceeds a threshold, wherein the removing results in the first input datum from the first training set not being used for a first training, which is an evaluation or judgement that can be performed in the human mind.
Thus, the claim recites an abstract idea under Step 2A Prong 1.
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
for the reduction of training data, which merely specifies the particular field of use or particular technological environment in which the abstract idea (comparing the aggregated vectors) is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
via a system comprising an encoder, as the performance of an abstract idea on a computer is not more than instructions to "apply it" on a computer, which by MPEP 2106.05(f) cannot integrate an abstract idea into a practical application;
the method comprising: a) receiving a first input datum from the first set of training data, which is merely an insignificant extra-solution activity of data gathering, which by MPEP 2106.05(g) cannot integrate an abstract idea into a practical application;
the first set of training data comprising non-labeled video, radar and / or lidar frames of sensor data recorded in a vehicle, which merely specifies the particular field of use or particular technological environment in which the abstract idea (comparing the aggregated vectors) is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
b) propagating the first input datum …, wherein one or more feature vectors are assigned to the input datum …, which is merely an insignificant extra-solution activity of data transfer, which by MPEP 2106.05(g) cannot integrate an abstract idea into a practical application;
d) performing steps a) through c) with a second input datum from the first set of training data…, which is merely an insignificant extra-solution activity of performing repetitive steps, which by MPEP 2106.05(g) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)), implementation on a computer (MPEP 2106.05(f)), and the activity of data gathering/performing repetitive steps(MPEP 2106.05(g)) cannot provide significantly more, as storing and retrieving information in memory (receiving… input datum from… training data) is well understood, routine, and conventional (MPEP 2106.05(d)(II)(iv)) and performing repetitive steps ( performing steps a) through c)…) is well understood, routine, and conventional (ii. Performing repetitive calculations, Flook, 437 U.S. at 594, 198 USPQ2d at 199 (recomputing or readjusting alarm limit values);), and the combination of additional elements does not provide an inventive concept. Thus, the claim is ineligible.
Claim 2, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the first set of training data comprises video, radar and / or lidar frames, which merely specifies the particular field of use or particular technological environment in which the abstract idea (comparing the aggregated vectors) is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 3, dependent upon Claim 2, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the video, radar and / or lidar frames of the first set of training data are temporal sequences of sensor data or sensor data which have been recorded during a journey of a vehicle or sensor data which have been artificially generated so that they simulate sensor data of a journey of a vehicle., which merely specifies the particular field of use or particular technological environment in which the abstract idea is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 4, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the first and second input datums of the first set of training data are temporally consecutive datums in the temporal sequence of the training data, which merely specifies the particular field of use or particular technological environment in which the abstract idea (comparing the aggregated vectors) is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 5, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the training data of the first set of training data is used to train an algorithm for highly automated or autonomous control of vehicles, as the performance of an abstract idea on a computer is not more than instructions to "apply it" on a computer, which by MPEP 2106.05(f) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 6, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the steps a) to f) are performed directly when recording or generating the training data of the first set of training data, which merely specifies the particular field of use or particular technological environment in which the abstract idea is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application;
and in step f) the first input datum is removed from the first set of training data when the threshold value of the measure of similarity is exceeded, which is merely an insignificant extra-solution activity of data transfer, which by MPEP 2106.05(g) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, the additional elements, alone or in combination, do not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)),and the activity of data transfer(MPEP 2106.05(g)) cannot provide significantly more, as receiving or transmitting data over a network is well understood, routine, and conventional (MPEP 2106.05(d)(II)(i)) and the combination of additional elements does not provide an inventive concept. Thus, the claim is ineligible.
Claim 7, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the steps a) to f) are performed prior to training or preprocessing with the training data of the first set of training data, which merely specifies the particular field of use or particular technological environment in which the abstract idea is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 8, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the aggregated vector is a histogram vector that assigns to each prototype feature vector an integer representing the respective assigned number of the respective prototype feature vector, which merely specifies the particular field of use or particular technological environment in which the abstract idea is to be performed, which by MPEP 2106.05(h) cannot integrate the abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because the particular field of use or particular technological environment (MPEP 2106.05(h)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 9, dependent upon Claim 1, further recites
the measure of similarity in step e) is determined via a cosine similarity, which is a mathematical concept.
Thus, the claim recites an abstract idea under Step 2A Prong 1.
Under Step 2A Prong 2 and 2B, the claim does not recite any new additional elements which could integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, the claim is ineligible.
Claim 10, dependent upon Claim 1, further recites
the measure of similarity in step e) comprises comparing the first, the second, and a third aggregated vector, wherein the third aggregated vector was generated using steps a) through c) with a third input datum from the first set of training data, which is an evaluation or judgement that can be performed in the human mind.
Thus, the claim recites an abstract idea under Step 2A Prong 1.
Under Step 2A Prong 2 and 2B, the claim does not recite any new additional elements which could integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, the claim is ineligible.
Claim 11, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the encoder has been trained as part of an autoencoder, as the performance of an abstract idea on a computer is not more than instructions to "apply it" on a computer, which by MPEP 2106.05(f) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 12, dependent upon Claim 1, further recites
the encoder comprises a first set of prototype feature vectors learned during the training of the autoencoder, which is an evaluation or judgement that can be performed in the human mind.
Thus, the claim recites an abstract idea under Step 2A Prong 1.
Under Step 2A Prong 2 and 2B, the claim does not recite any new additional elements which could integrate the abstract idea into a practical application or provide significantly more than the abstract idea itself. Thus, the claim is ineligible.
Claim 13, dependent upon Claim 1, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the encoder and / or the autoencoder are implemented via a neural network or a convolutional neural network, as the performance of an abstract idea on a computer is not more than instructions to "apply it" on a computer, which by MPEP 2106.05(f) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more. Thus, the claim is ineligible.
Claim 14 recites a non-transitory computer-readable storage medium, thus an article of manufacture, one of the four statutory categories of patentable subject matter. However, Claim 20 further recites comprising program code which, when executed, performs precisely the methods of Claim 1. Therefore, Step 2A Prong 1 analysis remains the same. As for Step 2A Prong 2 and Step 2B: performance on a computer cannot integrate an abstract idea into a practical application (Step 2A Prong 2) nor provide significantly more than the abstract idea itself (Step 2B) (MPEP 2106.05(f)), Claim 14 are rejected as subject-matter ineligible for reasons set forth in the rejections of Claim 1.
Claim 15 recites a computer system, comprising a processor, thus a machine, one of the four statutory categories of patentable subject matter. However, Claim 15 further recites configured to perform precisely the methods of Claim 1. Therefore, Step 2A Prong 1 analysis remains the same. As for Step 2A Prong 2 and Step 2B: performance on a computer cannot integrate an abstract idea into a practical application (Step 2A Prong 2) nor provide significantly more than the abstract idea itself (Step 2B) (MPEP 2106.05(f)). Claim 15 is rejected as subject-matter ineligible for reasons set forth in the rejections of Claim 1.
Claim 18, dependent upon Claim 11, further recites no additional abstract ideas. However:
Under Step 2A Prong 2, the claim does not include any additional elements which integrate the abstract idea into a practical application, since the additional elements consist of:
the encoder is trained prior to conducting the method for the reduction of training data, as the performance of an abstract idea on a computer is not more than instructions to "apply it" on a computer, which by MPEP 2106.05(f) cannot integrate an abstract idea into a practical application.
Thus, the claim is directed towards an abstract idea.
Further, under Step 2B, the additional element does not provide significantly more than the abstract idea itself, because implementation on a computer (MPEP 2106.05(f)) cannot provide significantly more. Thus, the claim is ineligible.
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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-7, 9, 10, 14, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Birodkar et al., “Semantic Redundancies in Image-Classification Datasets: The 10% You Don’t Need” (hereinafter Birodkar) in view of Waymo, “Waymo Open Dataset” (hereinafter Waymo).
Regarding Claim 1, Birodkar teaches:
A method for the reduction of training data, (Birodkar [p.1 2nd col 2nd ¶] We find that at least 10% of ImageNet and CIFAR-10 can be safely removed by a technique as simple as clustering. Particularly, we identify a certain subset of ImageNet and CIFAR-10 whose removal does not affect the test accuracy when the architecture is trained from scratch on the remaining subset)
via a system comprising an encoder…, and the encoder maps input data to prototype feature vectors of a set of prototype feature vectors, (Birodkar [p.2 2nd col last ¶] To find redundancies in datasets, we look at the semantic space of a pre-trained model trained on the full dataset. In our case, the semantic representation comes from the penultimate layer of a neural network) (Note: the layer of the neural network corresponds to the encoder)
the method comprising: a) receiving a first input datum from the first set of training data; (Birodkar
PNG
media_image1.png
341
508
media_image1.png
Greyscale
) (Note: for each example, the leftmost image (enclosed in a box) corresponds to the first input datum)
b) propagating the first input datum by the encoder, wherein one or more feature vectors are assigned to the input datum by the encoder, and depending on the assigned feature vectors, a defined set of prototype feature vectors is determined and assigned to the first input datum; (Birodkar [p.2 2nd col last ¶] To find redundancies in datasets, we look at the semantic space of a pre-trained model trained on the full dataset. In our case, the semantic representation comes from the penultimate layer of a neural network… [p.3 1st col 2nd to last ¶] The semantic representation obtained was a 64- dimensional vector.) (Note: The semantic representation from a layer of a neural network corresponds to the propagation of the input data by the encoder and the resulting assigned feature vectors. Each dimension in the semantic representation corresponds to each prototype feature vector.)
c) creating an aggregate vector for the first input datum; (Birodkar [p.2 2nd col last ¶] To find groups of points which are close by in the semantic space we use Agglomerative Clustering... Agglomerative Clustering assumes that each point starts out as its own cluster initially, and at each step, the pair of clusters which are closest according to the dissimilarity criterion are joined together. Given two images I1 and I2, whose latent representations are denoted by vectors x1 and x2.)(Note: x1 corresponds to the aggregate vector for the first input datum)
d) performing steps a) through c) with a second input datum from the first set of training data and creating a second aggregate vector for the second input datum; (Birodkar
PNG
media_image1.png
341
508
media_image1.png
Greyscale
[p.2 2nd col last ¶] To find groups of points which are close by in the semantic space we use Agglomerative Clustering... Agglomerative Clustering assumes that each point starts out as its own cluster initially, and at each step, the pair of clusters which are closest according to the dissimilarity criterion are joined together. Given two images I1 and I2, whose latent representations are denoted by vectors x1 and x2.) (Note: the image next to the box corresponds to the second input datum. X2 corresponds to the second aggregate vector.)
e) comparing at least the first and second aggregated vectors and determining a measure of similarity for the aggregated vectors; and f) removing the first input datum from the first set of training data when the determined measure of similarity exceeds a threshold, wherein the removing results in the first input datum from the first training set not being used for a first training. (Birodkar [p.3 1st col 2nd ¶] For Agglomerative Clustering, we process points belonging to each class independently. Since the dissimilarity is a pairwise measure, processing each class separately leads to faster computations. We run the clustering algorithm until there are k clusters left, where k is the size of the desired subset. We assume that points inside a cluster belong to the same redundant group of images. In each redundant group, we select the image whose representation is closest to the cluster center and discard the rest. Henceforth, we refer to this procedure as semantic space clustering or semantic clustering for brevity.) (Note: discarding clusters that are the farthest from the cluster center corresponds to removing the first input data that exceed the threshold.)
Birodkar does not teach, but Waymo further teaches:
wherein at least part of the training data forms a temporal sequence and is combined in a first set of training data (Waymo [p.1 last ¶] The Waymo Open Dataset currently contains lidar and camera data from 1,000 segments (20s each). We plan to continuously grow this dataset. Here is what is currently included: 1,000 segments of 20s each, collected at 10Hz (200,000 frames) in diverse geographies and conditions)
the first set of training data comprising non-labeled video, radar and / or lidar frames of sensor data recorded in a vehicle (Waymo [p.1 last ¶] The Waymo Open Dataset currently contains lidar and camera data from 1,000 segments (20s each). We plan to continuously grow this dataset. Here is what is currently included: 1,000 segments of 20s each, collected at 10Hz (200,000 frames) in diverse geographies and conditions)
Waymo and Birodkar are analogous to the present invention because both are from the same field of endeavor of utilization of datasets in neural network-based training. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Waymo’s video dataset into the redundancy reduction method of Birodkar. The motivation would be to “to aid the research community in making advancements in machine perception and self-driving technology” (Waymo [p.1]).
Regarding Claim 3, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 2. Birodkar does not teach, but Waymo further teaches:
The method according to claim 2, wherein the video, radar and / or lidar frames of the first set of training data are temporal sequences of sensor data or sensor data which have been recorded during a journey of a vehicle or sensor data which have been artificially generated so that they simulate sensor data of a journey of a vehicle. (Waymo [p.1 last ¶] The Waymo Open Dataset currently contains lidar and camera data from 1,000 segments (20s each). We plan to continuously grow this dataset. Here is what is currently included: 1,000 segments of 20s each, collected at 10Hz (200,000 frames) in diverse geographies and conditions)
Waymo and Birodkar are analogous to the present invention because both are from the same field of endeavor of utilization of video-based datasets. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Waymo’s video dataset into the redundancy reduction method of Birodkar. The motivation would be to “to aid the research community in making advancements in machine perception and self-driving technology” (Waymo [p.1]).
Regarding Claim 4, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar does not teach, but Waymo further teaches:
The method according to claim 1, wherein the first and second input datums of the first set of training data are temporally consecutive datums in the temporal sequence of the training data. (Waymo [p.1 last ¶] The Waymo Open Dataset currently contains lidar and camera data from 1,000 segments (20s each). We plan to continuously grow this dataset. Here is what is currently included: 1,000 segments of 20s each, collected at 10Hz (200,000 frames) in diverse geographies and conditions)
Regarding Claim 5, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar does not teach, but Waymo further teaches:
The method according to claim 1, wherein the training data of the first set of training data is used to train an algorithm for highly automated or autonomous control of vehicles. (Waymo [p.2 last ¶] While this dataset is not reflective of the full capabilities of our sensor system and is only a fraction of the data on which Waymo’s self-driving system is trained…)
Waymo and Birodkar are analogous to the present invention because both are from the same field of endeavor of utilization of datasets in neural network-based training. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Waymo’s video dataset into the redundancy reduction method of Birodkar. The motivation would be to “to aid the research community in making advancements in machine perception and self-driving technology” (Waymo [p.1]).
Regarding Claim 6, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
The method according to claim 1, wherein the steps a) to f) are performed directly when recording or generating the training data of the first set of training data, and in step f) the first input datum is removed from the first set of training data when the threshold value of the measure of similarity is exceeded. (Birodkar [p.3 1st col 2nd ¶] For Agglomerative Clustering, we process points belonging to each class independently. Since the dissimilarity is a pairwise measure, processing each class separately leads to faster computations. We run the clustering algorithm until there are k clusters left, where k is the size of the desired subset. We assume that points inside a cluster belong to the same redundant group of images. In each redundant group, we select the image whose representation is closest to the cluster center and discard the rest. Henceforth, we refer to this procedure as semantic space clustering or semantic clustering for brevity.) (Note: creating the clustering corresponds to generating the training data of the first set of training data. Semantic space clustering to discard the redundant data corresponds to directly performing the removal from the first set of training data, as removing the data causes a direct effect to the dataset.)
Regarding Claim 7, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
7. The method according to claim 1, wherein the steps a) to f) are performed prior to training or preprocessing with the training data of the first set of training data. (Birodkar [p.1 2nd col 2nd ¶] We find that at least 10% of ImageNet and CIFAR-10 can be safely removed by a technique as simple as clustering. Particularly, we identify a certain subset of ImageNet and CIFAR-10 whose removal does not affect the test accuracy when the architecture is trained from scratch on the remaining subset)
Regarding Claim 9, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
The method according to claim 1, wherein the measure of similarity in step e) is determined via a cosine similarity. (Birodkar [p.2 2nd col last ¶] Given two images I1 and I2, whose latent representations are denoted by vectors x1 and x2. We denote the dissimilarity between x1 and x2 by d(x1, x2) using the cosine angle between them as follows:
PNG
media_image2.png
89
368
media_image2.png
Greyscale
)
Regarding Claim 10, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
The method according to claim 1, wherein the measure of similarity in step e) comprises comparing the first, the second, and a third aggregated vector, wherein the third aggregated vector was generated using steps a) through c) with a third input datum from the first set of training data. (Birodkar
PNG
media_image3.png
660
977
media_image3.png
Greyscale
[p.4 2nd col last ¶] To determine the best choice of semantic representation from a pre-trained model, we run experiments after selecting the semantic representation from 3 different layers in the network. Figure 8 shows the results. Here “Start” denotes the semantic representation after the first Convolution layer is a ResNet, “Middle“ denotes the representation after the second residual block, and “End” denotes the output of the last average pooling layer. We see that the “End” layer’s semantic representation is able to find the largest redundancy.
PNG
media_image4.png
465
516
media_image4.png
Greyscale
) (Note: for each row, the second image’s semantic representation corresponds to the second aggregated vector; the third image’s semantic representation corresponds to the third aggregated vector.)
Regarding Claim 14, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
A computer program product, comprising program code which, when executed, performs the method of claim 1. (Birodkar [p.3 1st col 3rd ¶] We use the ResNet… architecture for all our experiments… For each dataset, we compare the performance after training on different random subsets to subsets found with semantic clustering.) (Note: Resnet is the program product that performs the method of claim 1)
Regarding Claim 15, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar further teaches:
A computer system, set up to perform the method of claim 1. (Birodkar [p.3 1st col 3rd ¶] We use the ResNet… architecture for all our experiments… For each dataset, we compare the performance after training on different random subsets to subsets found with semantic clustering.) (Note: Resnet is the computer system that performs the method of claim 1)
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Birodkar/Waymo, further in view of Passalis et al., “Learning Bag-of-Features Pooling for Deep Convolutional Neural Networks” (hereinafter Passalis).
Regarding Claim 8, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar/Waymo does not teach, but Passalis further teaches:
The method according to claim 1, wherein the aggregated vector is a histogram vector that assigns to each prototype feature vector an integer representing the respective assigned number of the respective prototype feature vector (Passalis [p.5756 1st col 2nd ¶] In this work, a BoF-inspired layer, that acts as a trainable quantization-based pooling layer, is used between the feature extraction layer and the fully connected layer to resolve these problems… First, a number of feature vectors are extracted from an image using a handcrafted feature extractor… The number of feature vectors might vary according to the type of feature extraction and the size of each image. Then, these feature vectors are quantized into a predefined number of bins, called codewords. Finally, a constant length histogram representation is extracted for each image by counting the number of feature vectors that were quantized into each bin.) (Note: each predefined bin corresponds to the respective prototype feature vector.)
Passalis and Birodkar/Waymo are analogous to the present invention because both are from the same field of endeavor of representing convolutional layers in a neural network. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement Passalis’s quantization method with Birodkar/Waymo’s method of identifying and reducing redundancies. The motivation would be to “provide significantly better scale-invariance, while reducing the size of the resulting network.” (Passalis [p.5756 1st col 1st ¶]).
Claims 11-13 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Birodkar/Waymo in view of Zhang et al., “Unsupervised object-level video summarization with online motion auto-encoder” (hereinafter Zhang).
Regarding Claim 11, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar/Waymo does not teach, but Zhang further teaches:
The method according to claim 1, wherein the encoder has been trained as part of an autoencoder. (Zhang [p.377 1st col 2nd ¶] Unsupervised online dictionary learning. We propose a novel online motion-AE model, which can mimic the online dictionary learning for memorizing past states of object motions by continuously updating a tailored recurrent auto-encoder network.
[p.379 1st col 4th ¶] LSTM Auto-Encoder. In LSTM auto-encoder, both the encoder network and the decoder network are built upon the LSTM units. Given an input object clip X = (x1, ··· , xT ), the LSTM encoder recurrently output hidden states (h1, ··· , hT ) with shared network parameters
PNG
media_image5.png
22
139
media_image5.png
Greyscale
, where WE is the transform from the input to LSTM encoder states.)
Zhang and Birodkar/Waymo are analogous to the present invention because both are from the same field of endeavor of disambiguating valuable video frames from a video dataset. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement Zhang’s autoencoder to Birodkar/Waymo’s redundancy reduction method. The motivation would be to “not only capture the detailed appearance of each moving object, but also its corresponding location and surrounding context” (Zhang [p.380 2nd col 1st ¶]).
Regarding Claim 12, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 11. Birodkar/Waymo does not teach, but Zhang further teaches:
The method according to claim 11, wherein the encoder comprises a first set of prototype feature vectors learned during the training of the autoencoder. (Zhang [p.378 1st col 3rd ¶] 3. Online motion Auto-Encoder (online motion-AE) The proposed online motion-AE framework resolves the finegrained unsupervised video summarization problem by training a stacked sparse LSTM auto-encoder in an online manner… [p.380 2nd col 1st ¶] … not only capture the detailed appearance of each moving object, but also its corresponding location and surrounding context.)
Zhang and Birodkar/Waymo are analogous to the present invention because both are from the same field of endeavor of disambiguating valuable video frames from a video dataset. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement Zhang’s method of generating feature vectors with an autoencoder to Birodkar/Waymo’s redundancy reduction method. The motivation would be to “not only capture the detailed appearance of each moving object, but also its corresponding location and surrounding context” (Zhang [p.380 2nd col 1st ¶]).
Regarding Claim 13, Birodkar/Waymo respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Birodkar/Waymo does not teach, but Zhang further teaches:
The method according to claim 1, wherein the encoder and / or the autoencoder are implemented via a neural network or a convolutional neural network. (Zhang [p.378 1st col 4th ¶] The key towards the fine-grained unsupervised object-level video summarization is the model capability of capturing key object-level motion clips within each video. Inspired by the success achieved by recurrent neural networks (especially Long Shot-Term Memory (LSTM) [16]) on sequential modeling, our core learning module is established by plugging the hierarchical three LSTM layers into a generative auto-encoder model with the sparse constraint, as illustrated in Fig. 4.)
Zhang and Birodkar/Waymo are analogous to the present invention because both are from the same field of endeavor of disambiguating valuable video frames from a video dataset. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement Zhang’s autoencoder to Birodkar/Waymo’s redundancy reduction method. The motivation would be to “not only capture the detailed appearance of each moving object, but also its corresponding location and surrounding context” (Zhang [p.380 2nd col 1st ¶]).
Regarding Claim 18, Birodkar/Waymo/Zhang respectively teaches and incorporates the claimed limitations and rejections of Claim 11. Birodkar, via Birodkar/Waymo/Zhang, further teaches:
The method according to claim 11, wherein the encoder is trained prior to conducting the method for the reduction of training data. (Birodkar [p.2 2nd col last ¶] To find redundancies in datasets, we look at the semantic space of a pre-trained model trained on the full dataset. In our case, the semantic representation comes from the penultimate layer of a neural network.)
Response to Arguments
Applicant’s arguments filed 06/09/2025 have been fully considered, but are not fully persuasive.
The objections to the claims and the specifications have been withdrawn in view of the amendments and/or the arguments. However, the amendment to claim 14 has raised a new objection to the Specification.
With respect to the rejection under 35 USC § 101:
The rejections to claims 14 and 15 regarding nonstatutory subject matter have been withdrawn in view of the amendments.
The rejection of claims 1-15 under 35 USC 101 as being directed to abstract idea without significantly more:
Applicant argues that the claims recite an improvement in a technology, specifically the data reduction in the training of the ANN in the instant application, as described in para. 5 and para 12 of the Specification.
In response, Examiner respectfully submits that the features (data reduction in the training of the ANN) are not positively recited in the claims, and thus the claims remain directed towards the judicial exceptions as set forth in the rejection. Even if the claims did positively recite the reduction in data, the improvement would be towards reducing data, which is an abstract idea and not a technology. Rather, the claims recite a neural network which is generally linked to the abstract idea recited in the claims; therefore, the claims neither recite nor reflect the technology of training the neural network. Thus, the claim as recited does not amount to an integration into a practical application of or significantly more than the judicial exception. The rejection is maintained.
With respect to the rejection under 35 USC § 103:
(p.10) Applicant asserts that Birodkar does not teach “the aggregated vectors of two temporally adjacent input datums…” Examiner respectfully submits that, under the broadest reasonable interpretation, there are no details to the threshold of the input data (i.e., no limit to the temporal adjacency). For this reason, Birodkar does teach a threshold for removing datums from the training data via the relative cluster centers.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., “the aggregated vectors of two temporally adjacent input datums…”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicant further asserts that the new claims 16-18 are allowable by the analogous reasons set forth above. Examiner respectfully submits the arguments are unpersuasive for the reasons set forth above in response to arguments regarding claim 1, and that the combination of Birodkar/Waymo/Zhang does teach claims 16-18 (Birodkar [p.2 2nd col last ¶], Zhang [p.377 1st col 1st ¶]; see p.32-34 of the Office Action).
Thus, claims 1-18 remain unpatentable.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOSEP HAN whose telephone number is (703)756-1346. The examiner can normally be reached Mon-Fri 9am-5pm.
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, Kakali Chaki can be reached on (571) 272-3719. 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.
/J.H./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights