DETAILED ACTION
This Action is responsive to Claims filed 06/25/2026.
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 .
Status of the Claims
Claims 1, 14, 18, and 20 have been amended. Claims 4 and 17 have been cancelled. Claims 21 and 22 are new. Claims 1-3, 5-16, and 18-22 are currently pending.
Response to Arguments
Applicant's arguments, see Pages 7-12, filed 06/25/2026, regarding the 35 U.S.C 101 Rejection of 14-19 have been fully considered but they are not persuasive.
The Applicant argues that the inclusion of the amended limitations in Claim 14 renders the claims eligible subject matter. The Examiner respectfully disagrees with the Applicant. While the additional elements bring the claim more in line with Claim 1, the majority of the amendments merely recite refinements to the structure of the discretization level reduction model already cited as a non-generic computing component generally linking the subsequent abstract idea mental process steps to a particular technological field. The “…based on…” amendments to the interpretable abstract idea mental process steps do not sufficiently preclude a human mind from performing the claimed steps of generically determining tensor data generically “based on” the reduction layers.
As presently drafted, the “determining…”, “determining…”, “determining…”, and “adjusting…” steps do not recite specific structure or implementation precluding the steps from being practically performed within the human mind or with the aid of pen and paper, or are not phrased in such a way as to be interpretable as additional elements, rather than abstract ideas (as Claim 1 is phrased).
Given this interpretation, the Examiner submits the specific improvement alleged by the Applicant is derived from the data observation/manipulation recited in the interpretable abstract idea mental process steps. Per MPEP 2106.05(a), the specific improvement cannot come from the abstract idea. See the updated 35 U.S.C. 101 Rejection below.
Applicant's arguments, see Pages 12-15, filed 06/25/2026, regarding the prior art Rejection under 35 U.S.C. 102(a)(1) and 103 have been fully considered but they are not persuasive.
The amendments made to the independent claims do not sufficiently differentiate the claims from the teachings of Hou. The Applicant argues Hou does not read on the “having a number…” limitation. The Examiner respectfully disagrees. Based on the claim language and the BRI of both a “tensor” and “discretization level” used in the claims, the Examiner submits Hou’s reduction of an array/tensor with fixed lengths of rows reads on “a discrete number of values that may be held at each position of the input tensor.” For example, if one were to take a 5 x 5 input array/tensor, then each of the 5 rows contains a discrete number of values (5) at each array/tensor position (0th row contains 5 values, 1st row contains 5 values, etc.). Hou’s network architecture and reduction of the input image through each reduction layer (Figure 2), reads on this interpretation, in the absence of further clarifying amendments.
Given that the Examiner contends the 102(a)(1) Rejection should be upheld, the 35 U.S.C. Rejection remains unchanged, as Claim 8 has not been amended.
See the updated 102(a)(1) and 103 Rejections below.
Claim Rejections - 35 USC § 101
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claims 14-19 and 21 rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more; and because the claims as a whole, considering all claim elements both individually and in combination, do not amount to significantly more than the abstract idea, see Alice Corporation Pty. Ltd. v. CLS Bank International, et al, 573 U.S. (2014). In determining whether the claims are subject matter eligible, the Examiner applies the 2019 USPTO Patent Eligibility Guidelines. (2019 Revised Patent Subject Matter Eligibility Guidance, 84 Fed. Reg. 50, Jan. 7, 2019.)
Step 1:
Claims 14-19 and 21 recite a method, which falls under the statutory category of a process.
Step 2A – Prong 1:
Claim 14 recites an abstract idea, law of nature, or natural phenomenon. The limitations “determining, by the computing system and based at least in part on the one or more level reduction layers of the discretization level reduction model, the level-reduced tensor data;”, “determining, by the computing system and based at least in part on the one or more level reduction layers of the discretization level reduction model, reconstructed input tensor data based at least in part on the level-reduced tensor data;”, “determining, by the computing system, a loss based at least in part on the input tensor data and the reconstructed input tensor data;”, and “adjusting, by the computing system, one or more parameters of the discretization level reduction model based at least in part on the loss.” under the broadest reasonable interpretation, cover a mental process including an observation, evaluation, judgment or opinion that could be performed in the human mind or with the aid of pencil and paper. These limitations therefore fall within the mental process group.
Determining tensor data is practically performed within the human mind or with the aid of pen and paper. Determining a reconstructed tensor is practically performed within the human mind or with the aid of pen and paper. Determining a loss is practically performed within the human mind or with the aid of pen and paper. Adjusting model parameters is practically performed within the human mind or with the aid of pen and paper.
Step 2A – Prong 2:
The additional elements of claim 1 do not integrate the abstract idea into a judicial exception. The claim recites the additional elements “A computer-implemented method”, “tensor data”, and “a computing system comprising one or more computing devices” are recognized as generic computer components recited at a high level of generality (the Specification does not indicate these elements are different from a typical processing unit). Although it has and executes instructions to perform the abstract idea itself, this also does not serve to integrate the abstract idea into a practical application as it merely amounts to instructions to "apply it." (See MPEP 2106.04(d)(2) indicating mere instructions to apply an abstract idea does not amount to integrating the abstract idea into a practical application).
The additional elements recited in the limitations “a discretization level reduction model to provide level-reduced tensor data”, “input layer”, “reconstruction layers”, and “reduction layers” are recognized as non-generic computer components, however, they are found to generally link the abstract idea to a particular technological field (See MPEP 2106.05(h)).
The limitations “obtaining, by a computing system comprising one or more computing devices, training data, the training data comprising input tensor data having a number of discretization levels, the discretization levels defining a discrete number of values that may be held at each tensor position of the input tensor data;” and “providing, by the computing system, the training data to a discretization level reduction model, the discretization level reduction model configured to receive tensor data comprising the number of discretization levels and produce, in response to receiving the tensor data, level- reduced tensor data comprising a reduced number of discretization levels,” are found to pre- or post-extra-solution activity or data transmittal steps (See MPEP 2106.05(g)).
Step 2B:
The only limitation on the performance of the described method is a limitation reciting “A computer-implemented method”, “tensor data”, and “a computing system comprising one or more computing devices” These elements are insufficient to transform a judicial exception to a patentable invention because the recited elements are considered insignificant extra-solution activity (generic computer system, processing resources, links the judicial exception to a particular, respective, technological environment). The claim thus recites computing components only at a high-level of generality such that it amounts to no more than mere instructions to apply the exception using generic computer components; mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (see MPEP 2106.05(f)).
The additional elements recited in the limitations “a discretization level reduction model to provide level-reduced tensor data”, “input layer”, “reconstruction layers”, and “reduction layers” are recognized as non-generic computer components, however, they are found to generally link the abstract idea to a particular technological field (See MPEP 2106.05(h)).
The limitations “obtaining, by a computing system comprising one or more computing devices, training data, the training data comprising input tensor data having a number of discretization levels, the discretization levels defining a discrete number of values that may be held at each tensor position of the input tensor data;” and “providing, by the computing system, the training data to a discretization level reduction model, the discretization level reduction model configured to receive tensor data comprising the number of discretization levels and produce, in response to receiving the tensor data, level- reduced tensor data comprising a reduced number of discretization levels,” are found to well-understood, routine, or conventional activity (See MPEP 2106.05(d)(II)(i)(first list)).
Taken alone or in ordered combination, these additional elements do not amount to significantly more than the above-identified abstract idea. There is no indication that the combination of elements improves the functioning of a computer or improves any other technology. Their collective functions merely provide conventional computer implementation.
Dependent Claims:
Claim 15 recites refinements to the data types operated upon.
Claim 16 recites refinements to the loss calculated in Claim 14.
Claim 18 recites refinements to the model additional element(s) of Claim 14, mere pre- or post-extra-solution activity steps “obtaining…” and “obtaining…”, and abstract ide amental process step “determining…”.
Claim 19 recites refinements to the data type of Claim 18.
Claim 21 recites refinements to the additional elements of Claim 14.
Claim Rejections - 35 USC § 102
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claim(s) 1-3, 5-7, 9-16, and 18-22 is/are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Hou et al. (Image Companding and Inverse Halftoning using Deep Convolutional Neural Networks, 2017), hereinafter Hou.
In regards to claim 1: The present invention claims: “A computer-implemented method for providing level-reduced tensor data having improved representation of information, the method comprising: obtaining input tensor data having a number of discretization levels, the discretization levels defining a discrete number of values that may be held at each tensor position of the input tensor data;” Hou Figure 2 shows a 3x256x256 input tensor, and teaches “In all the experiment we use 3 bit depths (8 color levels) as input images to train our deep networks for both color and grayscale images.” (Page 6). Based on the claim language and the BRI of both a “tensor” and “discretization level” used in the claims, the Examiner submits Hou’s reduction of an array/tensor with fixed lengths of rows reads on “a discrete number of values that may be held at each position of the input tensor.”
“providing the input tensor data as input to a machine-learned discretization level reduction model configured to receive tensor data comprising the number of discretization levels and produce, in response to receiving the tensor data, level-reduced tensor data comprising a reduced number of discretization levels,” The purple blocks of Hou’s Figure 2 show the input tensor data being reduced at each layer. Hou also teaches “The typical digital halftoning process is considered as a technique of converting a continuous-tone grayscale image with 255 color levels (8 bits) into a binary black-and-white image with only 0 and 1 two color levels (1 bit).” (Page 2) and “The details of our model are shown in Fig. 2, we first encode the input image to lower dimension vector by a series of stride convolutions, which consists of 4 x 4 convolution kernels and 2 x 2 stride in order to achieve its own downsampling.” (Page 4).
“wherein the machine-learned discretization level reduction model comprises: at least one input layer configured to receive the tensor data;” Hou Figure 2 shows a 3x256x256 input tensor and Hou teaches “The details of our model are shown in Fig. 2, we first encode the input image to lower dimension vector by a series of stride convolutions…” (Page 4).
“and one or more level reduction layers connected to the at least one input layer, the one or more level reduction layers respectively configured to receive input having the first number of discretization levels and to provide a layer output having a second number of discretization levels, the second number of discretization levels being reduced relative to the first number of discretization levels;” Hou Figure 2 shows each layer reducing its respective input tensor and teaches “The dash-line arrows indicate the features from the encoding layers are directly copied to the decoding layers and form half of the corresponding layers’ features.”
“wherein the discretization level is reduced at each level reduction layer of the one or more level reduction layers based at least in part on a discretized activation function having the second number of discretization levels associated with the level reduction layer;” Hou teaches “we use 8 bit images as our highest bit depth images in the experiments. The 8 bit images are reduced by different depths as the lower bit depth images, and then expanded back to 8 bits.” (Page 4)
“obtaining, from the machine-learned discretization level reduction model, the level- reduced tensor data;” Hou teaches “to firstly encode the input images through several convolutional layers until a bottleneck layer, followed by a reversed decoding process to produce the output images.” (Page 3) and “we use 8 bit images as our highest bit depth images in the experiments. The 8 bit images are reduced by different depths as the lower bit depth images, and then expanded back to 8 bits.” (Page 4).
“wherein the machine-learned discretization level reduction model is trained using reconstructed input tensor data generated using an output of the machine-learned discretization level reduction model.” Hou teaches “to firstly encode the input images through several convolutional layers until a bottleneck layer, followed by a reversed decoding process to produce the output images.” (Page 3) and “We denote the loss function as L(^y; y) to measure the perceptual difference between two images. As illustrated in Fig. 1, both the output image ^y = T(x) generated by the transformation network and the corresponding target image y are fed into a pretrained deep CNN _ for feature extraction. We use _i(y) to represent the hidden representations of image y at ith convolutional layer. _i(x) is a 3D array of shape [Ci, Wi, Hi], where Ci is the number of filters, Wi and Hi are the width and height of the given feature map of the ith convolutional layer. The final perceptual loss of two images at ith layer is the Euclidean distance of the corresponding 3D arrays as following…” (Page 4).
In regards to claim 2: The present invention claims: “wherein the input tensor data comprises image data, and wherein the level-reduced tensor data comprises binarized image data.” Hou teaches “The typical digital halftoning process is considered as a technique of converting a continuous-tone grayscale image with 255 color levels (8 bits) into a binary black-and-white image with only 0 and 1 two color levels (1 bit).” (Page 2).
In regards to claim 3: The present invention claims: “wherein the discretization level reduction model further comprises at least one feature representation layer configured to map the input tensor data from the input layer to a feature representation of the input tensor data.” Hou teaches “We not only use a deep CNN as a nonlinear transformation function to map a low bit depth image to a higher bit depth image or from a halftone image to a continuous tone image, but also employ another pre-trained deep CNN as a feature extractor or convolutional spatial filter to derive visually important features to construct the objective function for the training of the transformation neural network.” (Page 2).
In regards to claim 5: The present invention claims: “wherein the one or more level reduction layers are each configured to reduce the number of discretization levels based at least in part on a scaling factor.” Hou teaches “The default approach [3] for converting 8 bit images to 4 bit images is to divide by 16 to quantize the color level from 256 to 16, which will be then scaled up to fill the full range of the display.” (Page 4).
In regards to claim 6: The present invention claims: “wherein the scaling factor is one half.” See the rejection of Claim 5 and Hou Equation 2 (Page 4) for converting 8 bit images to 4 bit images (one half).
In regards to claim 7: The present invention claims: “wherein the one or more level reduction layers progressively and monotonically reduce a number of discretization levels at each of the one or more level reduction layers.” See Hou Figure 2 for each purple block reducing per layer.
In regards to claim 9: The present invention claims: “wherein the machine-learned discretization level reduction model comprises an output layer configured to provide the level-reduced tensor data.” Hou teaches “converting a continuous-tone grayscale image with 255 color levels (8 bits) into a binary black-and-white image with only 0 and 1 two color levels (1 bit).” (Page 2).
In regards to claim 10: The present invention claims: “wherein the reduced number of discretization levels of the level-reduced tensor data is two discretization levels.” See the above Rejection of Claim 9 how Hou reads on reducing input to two discretization levels.
In regards to claim 11: The present invention claims: “wherein the discretization level reduction model comprises one or more reconstruction layers configured to reconstruct the reconstructed input tensor data from the level-reduced tensor data.” See Hou Figure 2 for the orange blocks representing reconstruction layers. Hou teaches “By inverting convolutional features [39], the colors and the rough contours of an image can be reconstructed from activations in pretrained CNNs.” (Page 3).
In regards to claim 12: The present invention claims: “wherein the discretization level reduction model comprises a color bypass network, the color bypass network comprising one or more fully connected hidden units.” See Hou Figure 2 for “The dash-line arrows indicate the features from the encoding layers are directly copied to the decoding layers and form half of the corresponding layers’ features.” (Page 3).
In regards to claim 13: The present invention claims: “wherein the color bypass network comprises between one and ten fully connected hidden units.” See above and Hou Figure 2 for the multiple hidden layers convolving and deconvolving the input tensor.
In regards to claims 14-15, 18, and 21: Claims 14-15, 18, and 21 recite similar limitations to those found in Claims 1-13, save for the recitation of “A computer-implemented method for training a discretization level reduction model to provide level-reduced tensor data having improved representation of information, the computer-implemented method comprising:” of Claim 14; therefore, both claims are similarly rejected.
In regards to claim 16: The present invention claims: “wherein the loss comprises a pixel-wise difference between the input tensor data and the reconstructed input tensor data.” While Hou does not utilize pixel-wise difference loss directly (See Page 3 “Instead of using perpixel losses, i.e. measuring pixel-wise difference between the output image and its target (the original) image, we measure the difference between the output image and target image based on the high level features extracted from pretrained deep convolutional neural networks.”), Section II.C of Hou does go into detail regarding the state of the art of per-pixel difference loss at the time of Hou’s writing, as well as the deficiencies such a method has compared to the method used by their system. A person of ordinary skill in the art before the Applicant’s filing date would have been aware of per-pixel difference loss and its relevant benefits of deficiencies.
In regards to claim 19: The present invention claims: “wherein the first reconstructed input tensor data component comprises a reconstructed image and wherein the second reconstructed input tensor data component comprises a color tint for the reconstructed image.” Hou teaches “We not only use a deep CNN as a nonlinear transformation function to map a low bit depth image to a higher bit depth image or from a halftone image to a continuous tone image, but also employ another pre-trained deep CNN as a feature extractor or convolutional spatial filter to derive visually important features to construct the objective function for the training of the transformation neural network.” (Page 2) and “By inverting convolutional features [39], the colors and the rough contours of an image can be reconstructed from activations in pretrained CNNs.” (Page 3).
In regards to claims 20: Claim 20 recites limitations similar in scope to those found in Claims 1 and/or 7, save for the recitation of “One or more non-transitory, computer-readable media storing a machine-learned discretization level reduction model configured to…” in Claim 20; therefore, both Claims are similarly rejected.
In regards to claim 22: The present invention claims: “wherein the discretization level is associated with a bit depth of the input tensor data.” Hou teaches “In all the experiment we use 3 bit depths (8 color levels) as input images to train our deep networks for both color and grayscale images.” (Page 6).” and “we use 8 bit images as our highest bit depth images in the experiments. The 8 bit images are reduced by different depths as the lower bit depth images, and then expanded back to 8 bits.” (Page 4).
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
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.
Claim(s) 8 is/are rejected under 35 U.S.C. 103 as being unpatentable over Hou as applied to Claim 1 above, in further view of Banerjee et al. (An Empirical Study on Generalizations of the ReLU Activation Function, 2019), hereinafter Banerjee.
In regards to claim 8: Hou does not explicitly teach the use of tanh activation functions, instead using LeakyReLU or ReLU (Page 4), as is claimed in “wherein the discretized activation function is a discretized tanh function.” However; Banerjee teaches “However Linear, Sigmoid, Tanh and ReLU are the most commonly used activation functions and they are often selected empirically during the network design phase, rather than through a proper data driven process.” (Abstract) and “A variation of the Tanh activation, called the Leaky-Tanh breaks the symmetry of Tanh by penalizing the negative part [10] and has been shown to be better than ReLU and leaky-ReLU for training deep neural networks.” (introduction).
Banerjee illustrates the tanh and ReLU activation functions are commonly used and well-known in the art before the Applicant’s filing date. It would have been obvious to one of ordinary skill in the art at the time of the Applicant’s filing to use either activation function to fit their specific needs in a system similar to Hou’s.
Conclusion
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.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GRIFFIN T BEAN whose telephone number is (703)756-1473. The examiner can normally be reached M - F 7:30 - 4:30.
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, Li Zhen can be reached at (571) 272-3768. 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.
/GRIFFIN TANNER BEAN/ Examiner, Art Unit 2121
/Li B. Zhen/ Supervisory Patent Examiner, Art Unit 2121
Prosecution Insights