DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
1 This action is in response to the amendment filed on 08/12/2025. The specifications were amended to correct the statements stated by the previous rejection. Claims 1-15, 17, and 19 have been amended, and claims 16, 18, and 20 were cancelled. Claims 1-10, 13-15, 17, and 19 remain rejected, but claims 11 and 12 are objected.
Response to Arguments
2 Applicant’s arguments with respect to claims 1, 17, and 19 filed on 08/12/2025, with respect to the rejection under 35 U.S.C. § 103 regarding that the prior art does not teach “selecting a coding mode based on loss values of the SS7 message that are greater than a loss threshold value, where the coding mode corresponds to a lossless encoding scheme”. The applicant incorporated elements from claim 16 (and by extension claims 18 and 20) into claims 1, 17, and 19 respectively, and then cancelling claims 16, 18, and 20 on their own. This argument has been considered, but are moot due to new grounds of rejections. The prior art from Ahmadi (from claim 16 due to the new combination) has been dropped from the rejection in favor of the prior art taught by Tu based on the new amended claims.
3 Regarding arguments to claims 2-5, 8-15, they directly/indirectly depend on independent claims 1, 17, and 19 respectively. Applicant does not argue anything other than independent claims 1, 17, and 19. The limitations in those claims, in conjunction with combination, was previously established as explained, with the exceptions of claims 11 and 12 which are considered to be allowable with the new changes, but are currently objected due to being dependent on claim 1. In addition, claim 8 is under new grounds of rejection, with prior art from Q. Wang has been dropped from the rejection in favor of the prior art taught by Lu based on the new amended claim.
4 Regarding arguments to claims 6 and 7, not only is it directly/indirectly depend on independent claim 1 respectively, but also argued that the examiner failed to provide "articulated reasoning with some rationale underpinning to support the legal conclusion of obviousness". This argument has been considered, but are moot due to new grounds of rejections. The prior art from Wu and Hussain has been dropped from the rejection in favor of the prior art taught by Budagavi based on the new amended claims.
5 Claims 16, 18, and 20 have been cancelled by the applicant as mentioned previously, therefore they will not be reviewed further.
Claim Rejections - 35 USC § 103
6 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.
7 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.
8 Claim(s) 1-2, 9-10, 17, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter “Guarda”) in view of Yang et al. (US 20110058608 A1) and Tu et al. (US 20160261870 A1).
9 Regarding claim 1, Guarda teaches an electronic device, comprising: a memory configured to store a set of rate distortion (RD) operation points ([Section IV. A] reciting “Thus, to obtain multiple RD trade-offs, i.e. RD points offering different rate-quality trade-offs…”) and a set of coding modes ([Section III. B] reciting “This way, each PC block may be coded using one from several available DL models, which behave like coding modes as in traditional coding, chosen in order for ADL-PCC to maximize the compression performance for PCs with different characteristics, both globally (PC-level) and locally (block-level).”); and circuitry configured to:
receive a three-dimensional (3D) point cloud geometry ([FIG 1] reciting “Input PC”);
partition the 3D point cloud geometry into a set of blocks ([FIG 1] reciting “PC Block Partitioning”);
select a block from the set of blocks ([Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized.”);
compute, for the selected block, a set of loss values ([Section IV. A] reciting “The training loss function has a fundamental impact on the obtained DL coding models and, thus, on the final RD performance. The goal of the loss function is to minimize the distortion between the original and reconstructed blocks, while minimizing the associated block coding rate.”) associated with at least one of a set of compression metrics ([Section III. I] reciting “…reconstructed block with DL coding model i, DMSE is the mean squared error for the selected distortion metric…For the selected distortion and rate metrics, it was found after extensive testing that very low λmode values perform better, meaning that the distortion tends to dominate.”), wherein
the set of loss values corresponds to the set of coding modes and a first subset of the set of coding modes is associated with a subset of the set of RD operation points ([Abstract] reciting “…selects the most suitable available deep learning coding model to code each block, thus maximizing the compression performance.”);
select, from the first subset of the set of coding modes, a first coding mode based on a specific loss value of the set of loss values ([FIG 1] reciting “DL Coding Model Selection.”) that is ;
select, from the first subset of the set of coding modes, a second coding mode based on each of the set of loss values ([Section III. B] reciting “This way, each PC block may be coded using one from several available DL models, which behave like coding modes as in traditional coding, chosen in order for ADL-PCC to maximize the compression performance for PCs with different characteristics, both globally (PC-level) and locally (block-level).”) that is , wherein the second coding mode corresponds to ;
and encode the selected block based on one of the first coding mode or the second coding mode ([Section III. B] reciting “The reconstructed block for each DL coding model is evaluated to select the model which produces the best compression performance…”), wherein the second coding mode is different from the first coding mode ([Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized.”).
10 Guarda does not explicitly teach to select, from the first subset of the set of coding modes, a first coding mode based on a specific loss value of the set of loss values that is less than a loss threshold value; select, from the first subset of the set of coding modes, a second coding mode based on each of the set of loss values that is greater than the loss threshold value, wherein the second coding mode corresponds to a lossless encoding scheme.
11 Yang teaches select, from the first subset of the set of coding modes, a first coding mode based on a specific loss value of the set of loss values that is less than a loss threshold value ([0066] reciting “…(d3) if so (i.e. at least one estimated coding cost is below a threshold), selecting an intra coding mode for the current MB, encoding the MB using the selected encoding mode and reconstructing the encoded MB…”).
12 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda) to incorporate the teachings of Yang to provide a way to determine a value that is less than or below a loss threshold for a specific coding mode or model as stated by Guarda. Doing so would reduce the coding cost of a certain threshold(s), as well as to reduce the threshold as stated by Yang ([0080] recited).
13 Guarda in view of Yang does not explicitly teach to select, from the first subset of the set of coding modes, a second coding mode based on each of the set of loss values that is greater than the loss threshold value, wherein the second coding mode corresponds to a lossless encoding scheme.
14 Tu teaches to select, from the first subset of the set of coding modes, a second coding mode based on each of the set of loss values that is greater than the loss threshold value, wherein the second coding mode corresponds to a lossless encoding scheme ([0153] reciting “The techniques described above with respect to CU partitioning scheme selection and mode selection may enable video encoding device 20 to determine whether to partition the largest CU and select the coding mode of sub-blocks which correspond to the zeroth (e.g., root) and first depths in the quad-tree. Video encode 20 may implement a similar or the same procedure for larger depths with respect to the quad tree partitioning scheme. Alternatively, video encoder 20 may apply a simplified procedure for larger depths. For example, if the distortion of the block in the first depth is larger than a threshold, video encoder 20 may perform further partitioning.”; [0155] reciting “…show an improved coding efficiency with 7-12% encoding speed loss.”).
15 It would have been obvious to one with ordinary skill before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang) to incorporate the teachings of Tu to provide a method that can obtain a specific coding mode that is greater than a loss threshold value as well as to have the coding modes to correspond or to connect with a type of lossless encoding scheme, using the coding modes taught by Guarda in view of Yang. Doing so would allow fast encoding and partitioning techniques with improved visual quality as stated by Tu ([0153] recited).
16 Regarding claim 2, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), wherein the set of compression metrics includes at least one of a rate metric or a mean square error (MSE) metric (Guarda; [Section III. I] reciting “…where inPC is the input block, recPCi is the corresponding reconstructed block with DL coding model i, DMSE is the mean squared error for the selected distortion metric… For the selected distortion and rate metrics, it was found after extensive testing that very low λmode values perform better, meaning that the distortion tends to dominate.”).
17 Regarding claim 9, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), wherein each of the set of coding modes corresponds to a respective deep neural network of a plurality of deep neural networks (Guarda; [Abstract] reciting “…selects the most suitable available deep learning coding model to code each block, thus maximizing the compression performance.”; [Section 1] reciting “Deep learning (DL)”; FIG 1. (a); [Section III. B] reciting “The reconstructed block for each DL coding model is evaluated to select the model which produces the best compression performance…”), and
each of which is trained (Guarda; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized.”) the circuitry is further configured to:
train the plurality of deep neural networks (Guarda; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized.”), and encode, based on the trained plurality of deep neural networks, the selected block to generate an encoded block (Guarda; [Section III. B] reciting “DL-based block encoding and reconstruction - Each block is separately coded and reconstructed with the several available trained DL coding models…”).
18 Regarding claim 10, Guarda in view of Yang and Tu teaches the electronic device according to claim 9 (see claims 1 and 9 rejections above), wherein the circuitry is further configured to encode the selected block based on an application of the plurality of deep neural networks on the selected block (Guarda; [Section III. B] reciting “The key coding component is an autoencoder (AE), which learns a signal adaptive transform, followed by discretization and entropy encoding the resulting latents. A variational autoencoder (VAE) is used in addition to estimate the parameters used for entropy coding [21]. All these modules are trained together in an end-to-end fashion to learn the so-called DL coding model… The reconstructed block for each DL coding model is evaluated to select the model which produces the best compression performance…”), and
the specific deep neural network corresponds to the first coding mode ([Section I] reciting “Deep learning (DL)”; [FIG 1] reciting “Block 1” and “Adaptive DL-based PC coding”).
19 Claims 17 and 19 has similar limitations as of Claim 1, therefore it is rejected under the same rationale as Claim 1.
20 Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, further in view of Jia-Ching Wang et al. (US 20210224647 A1).
21 Regarding claim 3, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), wherein the circuitry is further configured to;
input the selected block ; and
.
22 Guarda in view of Yang and Tu does not explicitly teach to input the selected block to a classifier model; and compute the set of loss values based on an output of the classifier model.
23 Jia-Ching Wang teaches to input the selected block to a classifier model ([Claim 1] reciting “…wherein the neural network model comprises a Convolutional Neural Network (CNN)…”); and compute the set of loss values based on an output of the classifier model ([Claim 1] reciting “…the CNN so that each of the feature extractors individually generates a first feature block for each of the first training data and so that the first classifier generates a first classification result for each of the first training data, wherein the processor generates a first vector for each of the first training data based on the corresponding first feature blocks, and the domain discriminator generates a first domain discrimination result for each of the first training data according to the corresponding first vector, wherein the processor further calculates a first classification loss value according to a first classification label and the corresponding first classification result of each of the first training data belonging to the first domain”).
24 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang and Tu) to incorporate the teachings of Jia-Ching Wang to provide a classifier model instead of Guarda’s autoencoder while also being able to output loss values. Doing so would make transferring various blocks in a more accurate and improve state as stated by Jia-Ching Wang ([0064] recited).
25 Claim(s) 4-5 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), Tu et al. (US 20160261870 A1), and Jia-Ching Wang et al. (US 20210224647 A1) as of claims 1 and 3, further in view of Shrivastava et al. (US 20240220681 A1).
26 Regarding claim 4, Guarda in view of Yang, Tu, and Jia-Ching Wang teaches the electronic device according to claim 3 (see claims 1 and 3 rejections above), wherein
the circuitry is further configured to train the classifier model based on a plurality of geometric characteristics of a plurality of test blocks (Guarda; [Section III. B] reciting “In fact, due to massive differences in PCs characteristics, e.g. density, a single DL coding model may not be efficient for every PC (3D) block.”), and
the plurality of test blocks is associated with a point cloud (Guarda; [Section III. I] reciting “The overall rate used for signaling the selected models is negligible (less than 0.5% of the total PC rate, in the worst case for the selected test PCs).”).
27 Guarda in view of Yang, Tu, and Jia-Ching Wang does not explicitly teach the classifier model is a Deep Neural Network (DNN) model.
28 Shrivastava teaches the classifier model is a Deep Neural Network (DNN) model ([0066] reciting “At block 508, the process 500 can include combining a deep neural network (DNN) noise model (e.g., DNN noise model 408) with the generated point cloud from block 506.”).
29 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang, Tu, and Jia-Ching Wang) to incorporate the teachings of Shrivastava to provide the replacement of Guarda’s autoencoder for a Deep Neural Network (or a replacement of Jia-Ching Wang’s Convolutional Neural Network (CNN)) that was trained of various test blocks for a point cloud. Doing so would train the model to an accurate representation based on the specific setting as stated by Shivastava ([0066] recited).
30 Regarding claim 5, Guarda in view of Yang, Tu, Jia-Ching Wang, and Shrivastava teaches the electronic device according to claim 4 (see claims 1 and 3-4 rejections above), wherein the plurality of geometric characteristics comprises a density of points that is associated with the point cloud (Guarda; [Section III. B] reciting “In fact, due to massive differences in PCs characteristics, e.g. density, a single DL coding model may not be efficient for every PC (3D) block.”).
31 Claim(s) 6-7 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, and further in view of Budagavi et al. (US 20190139266 A1) and Vosoughi et al. (US 20200013215 A1).
32 Regarding claim 6, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above) wherein the circuitry is further configured to:
determine a portion of the 3D point cloud geometry (Guarda; [FIG 1] reciting “Input PC”) as a , wherein the portion of the 3D point cloud geometry corresponds to a subset of the set of blocks (Guarda; [Section III. B] reciting “The main novelty of this paper is the extension of a 2D DL-based coding architecture [21] to 3D PCs and its usage in an adaptive way to match the local PC characteristics… The complete PC is first partitioned into regular sized 3D blocks of voxels.”);
and encode the subset of the set of blocks (Guarda; [Section III. B] reciting “The first step of the encoding procedure (as represented in Fig. 1b) for each 3D block is to apply an AE that operates as a non-linear transform.”)
33 Guarda in view of Yang and Tu does not explicitly teach to determine a portion of the 3D point cloud geometry as a region of interest (ROI), and encode the subset of the set of blocks based on the lossless encoding scheme.
34 Budagavi teaches to determine a portion of the 3D point cloud geometry as a region of interest (ROI) ([0099] reciting “A ROI can indicate a portion of the point cloud that is modified differently than the rest of the 3-D point cloud 502.”).
35 It would have been obvious to one with ordinary skill before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang and Tu) to incorporate the teachings of Budagavi to provide a method to utilize a region of interest (ROI) with the three-dimensional point cloud methods taught by Guarda in view of Yang and Tu. Doing so would allow many types of metadata such as the scale, offset, rotation, point size, and point shape, which can modify how the decoder reconstructs the 3D point cloud as stated by Budagavi ([0159] recited).
36 Guarda in view of Yang, Tu, and Budagavi does not explicitly teach to encode the subset of the set of blocks based on the lossless encoding scheme.
37 Vosoughi teaches to encode the subset of the set of blocks based on the lossless encoding scheme ([0030] reciting “The electronic apparatus 102 may be further configured to encode the generated plurality of quantized levels. The electronic apparatus 102 may be configured to encode the plurality of quantized levels by entropy coding of the 3D point cloud. The entropy coding of the point cloud may result is a lossy compression of the 3D point cloud.”; [0039] reciting “The encoder circuitry 206 may be configured to generate a bit-stream of coded voxel attributes by application of an encoding scheme on the 3D point cloud. The bit-stream of coded voxel attributes may represent the attribute information corresponding to the 3D point cloud of the object.”).
38 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang, Tu, and Budagavi) to incorporate the teachings of Vosoughi the existing 3d point cloud geometry methods taught by Guarda in view of Yang, Tu, and Budagavi to have the same point cloud containing various loss function parameters as an encoding scheme. Doing so would allow the specific encoding scheme to comprise of suitable logic and circuity as stated by Vosoughi ([0039] recited).
39 Regarding claim 7, Guarda in view of Yang, Tu, Budagavi, and Vosoughi teaches the electronic device according to claim 6 (see claims 1 and 6 rejections above), wherein the circuitry is further configured to determine the portion of the 3D point cloud geometry as the ROI (Budagavi; [0099] reciting “A ROI can indicate a portion of the point cloud that is modified differently than the rest of the 3-D point cloud 502.”) based on at least one of a user input, an object detection operation, or a semantic segmentation operation (Budagavi; [0047] reciting “The display device 212A may comprise suitable logic, circuitry, interfaces, and/or code that may be configured to render the 3D point cloud onto a display screen of the display device 212A. In accordance with an embodiment, the display device 212A may include a touch screen to receive the user input.”).
40 Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, further in view of Lu et al. (US 20080008242 A1).
41 Regarding claim 8, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), wherein the each RD operation point of the set of RD operation points (Guarda; [Section IV. A] reciting “Thus, to obtain multiple RD trade-offs, i.e. RD points offering different rate-quality trade-offs…”)
42 Guarda in view of Yang and Tu does not explicitly teach wherein the each RD operation point of the set of RD operation points is associated with at least one of a plurality of loss threshold values, the at least one of the plurality of loss threshold values corresponds to at least one of the set of coding modes, and the plurality of loss threshold values includes the loss threshold value.
43 Lu teaches wherein the each RD operation point of the set of RD operation points is associated with at least one of a plurality of loss threshold values, the at least one of the plurality of loss threshold values corresponds to at least one of the set of coding modes, and the plurality of loss threshold values includes the loss threshold value ([0035] reciting “The video encoder includes an encoder for performing a multi-stage mode selection when encoding a current macroblock in the B slice, using a plurality of trained thresholds with multiple video sequences and fitted to a model that is linear with respect to .lamda..sub.MODE, where .lamda..sub.MODE is quantization parameter dependent, and comparing a Rate-Distortion (RD) cost to the plurality of trained thresholds when coding in DIRECT mode, and terminating the mode selection at different stages based on the RD cost.”; [0061] reciting “The threshold is set to a smaller value when high coding speed is preferred. On the other hand, to design a coder with little loss, the threshold can be set to the number of all available neighboring blocks.”).
44 It would have been obvious to one with ordinary skill before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang and Tu) to incorporate the teachings of Lu to provide a method to provide a plurality of a type of loss threshold values that can respond to a coding mode (like a direct mode) using the RD and coding mode methods provided by Guarda in view of Yang and Tu. Doing so would allow the threshold to be set to the number of all available neighboring blocks as stated by Lu ([0061] recited).
45 Claim(s) 13 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, and further in view of Neumann et al. (US 20220058506 A1).
46 Regarding claim 13, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above) wherein the circuitry is further configured to:
acquire a calibration point cloud; compute, for a block of the calibration point cloud (Guarda; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized.”), each coding mode of the set of coding modes of an RD operation point of the set of RD operation points (Guarda; [Abstract] reciting “…selects the most suitable available deep learning coding model to code each block, thus maximizing the compression performance.”; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized. This way, each PC block may be coded using one from several available DL models, which behave like coding modes as in traditional coding…”); and
to the first coding mode (Guarda; [Section III. H] reciting “The fixed entropy model is learned and optimized during the DL coding model training process, and used for all blocks to be coded.”)
47 Guarda in view of Yang and Tu does not explicitly teach a quartile of loss value corresponding each coding mode of the set of coding modes of an RD operation point of the set of RD operation points; and set the quartile of loss value corresponding to the first coding mode as the loss threshold value.
48 Neumann teaches a quartile of loss value corresponding each coding mode of the set of coding modes of an RD operation point of the set of RD operation points; and set the quartile of loss value corresponding to the first coding mode as the loss threshold value ([0039] reciting “Preconfigured threshold may be set by one or more expert users and/or determined statistically, for instance by finding a top quartile and/or number of percentiles of proximity in a series of distance determinations over time for user, at one time for a plurality of users, and/or over time for a plurality of users.”).
49 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda, Yang, and Tu) to incorporate the teachings of Neumann to provide a first quartile of a loss value that corresponds to Guarda’s RD operation points instead of Neumann’s distance determinations, as well as setting the value as the loss threshold. Doing so would find the top quartile of percentiles of proximity over time for users as stated by Neumann ([0039] recited).
50 Claim(s) 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, and further in view of He et al. (US 20220284246 A1).
51 Regarding claim 14, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), associated with each coding mode of the set of coding modes (Guarda; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized. This way, each PC block may be coded using one from several available DL models, which behave like coding modes as in traditional coding…”), and
the each coding mode of the set of coding modes corresponds to a respective RD operation point of the set of RD operation points (Guarda; [Section III. B] reciting “By training multiple DL coding models with each one targeting different PC characteristics, e.g. by adjusting the training loss function parameters, the usage of different DL models for each block type may be optimized. This way, each PC block may be coded using one from several available DL models, which behave like coding modes as in traditional coding…”; [Section III. H] reciting “The fixed entropy model is learned and optimized during the DL coding model training process, and used for all blocks to be coded.”; [Section IV. A] reciting “Thus, to obtain multiple RD trade-offs, i.e. RD points offering different rate-quality trade-offs…”).
52 Guarda in view of Yang and Tu does not explicitly teach wherein the loss threshold value is a specific value associated with each coding mode of the set of coding modes.
53 He teaches wherein the loss threshold value is a specific value associated with each coding mode of the set of coding modes ([0032] reciting “In a typical calculation of the loss function based on the hard margin with a fixed value, a total loss function is calculated only based on the hard margin loss function.”).
54 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang, and Tu) to incorporate the teachings of He to provide a fixed value, which can be specific, for a loss threshold that can correspond to the RD operation points taught by Guarda in view of Yang, and Tu. Doing so would avoid any problems for any retrieval effects of various models as stated by He ([0033] recited).
55 Claim(s) 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Adaptive Deep Learning-Based Point Cloud Geometry Coding," in IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 2, pp. 415-430, Feb. 2021, doi: 10.1109/JSTSP.2020.3047520. (hereinafter Guarda), Yang et al. (US 20110058608 A1), and Tu et al. (US 20160261870 A1) as of claim 1, and further in view of Jun Wang et al. (US 11232156 B1).
56 Regarding claim 15, Guarda in view of Yang and Tu teaches the electronic device according to claim 1 (see claim 1 rejection above), but does not explicitly teach wherein the circuitry is further configured to set the loss threshold value associated with the first coding mode based on a user input.
57 Jun Wang teaches to set the loss threshold value associated with the first coding mode based on a user input ([Col. 23; Lines 53-56] reciting “In an embodiment, the processor 204 may be configured to receive a user input, via the I/O device 210 from the user 116, to set the threshold value.”).
58 It would have been obvious to one with ordinary skill, before the effective filing date of the claimed invention, to have modified the method (taught by Guarda in view of Yang and Tu) to incorporate the teachings of Jun Wang to incorporate a loss threshold taught by Guarda in view of Yang and Tu that can be based on a user input. Doing so would allow the user to eliminate noisy blocks as stated by Jun Wang ([Col. 24; Lines 6-7] recited).
Allowable Subject Matter
59 Claims 11 and 12 are 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.
60 The following is a statement of reasons for the indication of allowable subject matter:
The limitations of Claim 11, which are the electronic device according to claim 1, wherein the set of loss values includes a first subset of loss values, the circuitry is further configured to compute the first subset of loss values for a second subset of the set of coding modes, the first subset of loss values corresponds to the second subset of the set of coding modes, and the second subset of the set of coding modes corresponds to a first RD operation point of the set of RD operation points, does not explicitly match with any prior art, or match with a combination of prior arts.
The limitations of Claim 12, which are the electronic device according to claim 11, wherein the circuitry is further configured to; switch to a second RD operation point of the set of RD operation points, based on the first subset of loss values that is greater than the loss threshold value for the second subset of the set of coding modes; and compute the set of loss values that includes a second subset of loss values for a third subset of the set of coding modest wherein the third subset of the set of coding modes corresponds to the second RD operation points the second RD operation point is different from the first RD operation point, the second subset of loss values is different from the first subset of loss values, and the first subset of the set of coding modes, the second subset of the set of coding modes, and the third subset of the set of coding modes are different, does not explicitly match with any prior art, or match with a combination of prior arts.
Conclusion
61 Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
62 Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOHNNY TRAN LE whose telephone number is (571)272-5680. The examiner can normally be reached Mon-Thu: 7:30am-5pm; First Fridays Off; Second Fridays: 7:30am-4pm.
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, Kent Chang can be reached at (571) 272-7667. 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.
/JOHNNY T LE/ Examiner, Art Unit 2614
/KENT W CHANG/ Supervisory Patent Examiner, Art Unit 2614
Prosecution Insights