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 .
Priority
Applicant’s claim for the benefit of prior-filed application 63/599541 filed 11/15/2023 under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged.
Information Disclosure Statement
The information disclosure statements (IDSs) submitted on 10/18/2024, 02/20/2025, 06/05/2025, 12/08/2025, 05/11/2026 are being considered by the examiner.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 20 recites the limitation "the quantization process" in line 2. There is insufficient antecedent basis for this limitation in the claim. Claim 1, from which Claim 20 depends indirectly recites only “an inverse quantization process.” Claim 13 and claims depending on Claim 13 recite “a quantization process.”
Claims 1-11 and 20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being incomplete for omitting essential steps, such omission amounting to a gap between the steps. See MPEP § 2172.01. The omitted steps are: a step connecting “receiving a time step parameter that is generated based on the latent representation” and “performing an inverse quantization process to generate a reconstructed latent representation.” That is, it is not clear what the inverse quantization process is performed on; previous steps recite “a quantized latent representation” and “a time step parameter,” but the claim does not explicitly recite the inputs to the inverse quantization process. Dependent Claims 2-11 and 20 do not remedy the omitted step; although Claim 2 recites the quantized latent representation is “entropy decoded before performing the inverse quantization process,” the limitation merely recites a sequence of events and does not explicitly recite the object(s) of the inverse quantization process action.
Claims 9-11 and 19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being incomplete for omitting essential steps, such omission amounting to a gap between the steps. See MPEP § 2172.01. The omitted steps are: a step connecting the network training and the generation of a time step parameter in Claims 9 and 19, and in Claim 11 a step connecting the network training and the generation of a quantization setting. Specification paragraph [0033] discloses “Because the optimal number of denoising time steps depends on the amount of noise in the latent representation and therefore the severity of quantization, and vice versa, parameter estimation network 120 predicts time step t and the quantization setting ϒ jointly. Parameter estimation network 120 may be trained to map the amount of noise in the latent representation y and the input setting λ to the optimal number of time steps t and the quantization setting ϒ.” Applicant is advised to incorporate elements of paragraph [0033] of the Specification, or otherwise clarify the training method of the network of Claims 9-11. As recited, there is a gap between the training and generating steps that are essential to the invention; it is not clear how the network obtains a time step parameter and a quantization setting.
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 1-8, 12-18 rejected under 35 U.S.C. 103 as being unpatentable over Yang (WO 2022166462 A1) in view of Ho et al. (Denoising Diffusion Probabilistic Models, published 2020).
Regarding Claim 1, Yang teaches “A method comprising:
receiving a quantized latent representation of an image in a latent space, wherein the image is encoded into a latent representation in the latent space and quantized to generate the quantized latent representation” (Yang, page 38, paragraph 1 discloses “Figure 6 shows a typical deep neural network-based image encoding and decoding scheme, which adopts an autoencoder network structure. The input x is the original image to be encoded, which can be expressed as an array of wx × hx × cx , where wx , hx , and cx represent the width, height, and number of color components of the input image, respectively. The analyzer performs a dimensionality reduction operation on the input image x to obtain its latent representation y…. Each element in the latent layer representation y output by the analyzer can be a floating-point number or an integer number, which can be quantized or ordinary rounded to obtain a more compact integer representation y'.” Yang, page 38, paragraph 2 also discloses “The quantization unit 208 in FIG. 2 is configured to perform a quantization operation on the transform coefficient 207 output by the transform processing unit 206 by applying scalar quantization or vector quantization to obtain a quantization level (quantization level) 209 of the transform coefficient 207, wherein the quantization level 209 It is the output of the quantized transform coefficient, also called the quantized transform coefficient 209, or the quantized coefficient 209”; where quantized transform coeffient is a quantized latent representation of an image)
performing an inverse quantization process to generate a reconstructed latent representation” (Yang, page 27, paragraphs 2-3 disclose “The inverse quantization unit 210 is configured to perform inverse quantization of the quantization unit 208 on the quantized coefficients 209 to obtain the dequantized coefficients 211, for example, performing the inverse of the quantization scheme performed by the quantization unit 208 according to or using the same quantization step size as the quantization unit 208” and “Inverse quantization unit 310 may be operable to receive quantization parameters (QPs) (or information related to inverse quantization in general) and quantization coefficients from encoded image data 21 (eg, parsed and/or decoded by entropy decoding unit 304)” and “The quantization parameters inverse quantize the decoded quantized coefficients 309 to obtain inverse quantized coefficients 311 , which may also be referred to as transform coefficients 311 or dequantized coefficients 311 “; where dequantized coefficients are a reconstructed latent representation);
decoding the (Yang, page 38, paragraph 1 discloses “Entropy decoding is the inverse operation of entropy encoding, and the purpose is to obtain y' from the compressed code stream.” Yang also discloses “decoder 30 includes entropy decoding unit 304, inverse quantization unit 310, inverse transform processing unit 312, reconstruction unit 314 (eg, summer 314), loop filter 320, decoded image buffer (DBP) ) 330 , a mode application unit 360 , an inter prediction unit 344 and an intra prediction unit 354 “ and “The reconstruction unit 314 (eg, summer 314) is used to add the reconstructed residual block 313 to the prediction block 365 to obtain the reconstructed block 315 in the pixel domain, for example, the pixel point values of the reconstructed residual block 313 and the prediction block 365 pixel values are added”; where reconstructed block in the pixel domain is a reconstructed image).
PNG
media_image1.png
623
611
media_image1.png
Greyscale
Figure 6 of Yang, machine translation
Yang does not explicitly teach “receiving a time step parameter that is generated based on the latent representation” and “performing, using a diffusion model, a denoising process for a number of iterations based on the time step parameter to remove noise from the reconstructed latent representation to generate a denoised reconstructed latent representation.”
However, Ho teaches “receiving a time step parameter that is generated based on the latent representation” (Ho Section 2, paragraph 1 discloses
PNG
media_image2.png
195
755
media_image2.png
Greyscale
Ho, Section 3.2 discloses “First, we set Σθ(xt ,t)=σ2tI to untrained time dependent constants”; where untrained time dependent constants are time step parameters generated based on latent representation)
and “performing, using a diffusion model, a denoising process for a number of iterations based on the time step parameter to remove noise from the reconstructed latent representation to generate a denoised reconstructed latent representation” (Ho, Section 3.4, paragraph 2 discloses “our diffusion process setup in Section 4 causes the simplified objective to down-weight loss terms corresponding to small t. These terms train the network to denoise data with very small amounts of noise, so it is beneficial to down-weight them so that the network can focus on more difficult denoising tasks at larger t terms”; where denoising data by a diffusion process is performing a denoising process using a diffusion model; see Figure 2, below which shows a diffusion model over iterations 1<t≤T).
PNG
media_image3.png
141
601
media_image3.png
Greyscale
Figure 2 of Ho
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yang to incorporate the teachings of Ho by implementing a diffusion model to perform denoising iteratively. The prior art Yang contained a ‘base’ method upon which the claimed invention can be seen as an ‘improvement.’ Yang teaches an encoder/decoder scheme for video or image compression. The claimed invention recites an encoder/decoder scheme for image compression, and also discloses a diffusion model performing denoising over a number of timesteps before decoding. The prior art contained a known technique that is applicable to the base method. Ho teaches a diffusion model that performs denoising on latent variables: Ho, Section 2 discloses “Diffusion models [53] are latent variable models,” and Ho, Section 3.4, paragraph 2 discloses “our diffusion process setup in Section 4 causes the simplified objective to down-weight loss terms corresponding to small t. These terms train the network to denoise data with very small amounts of noise, so it is beneficial to down-weight them so that the network can focus on more difficult denoising tasks at larger t terms.” One of ordinary skill in the art would have recognized that applying the known technique would have yielded predictable results and resulted in an improved system. That is, it would have been obvious to one of ordinary skill in the art that applying a denoising method designed for latent representations, as taught by Ho, to latent representations of images of Yang would have yielded predictable results of denoising images in the latent space. Additionally, Yang discloses denoising as a preprocessing step: Yang, page 20, paragraph 4 discloses “For example, the preprocessing performed by the preprocessor 18 may include trimming, color format conversion (eg, from RGB to YCbCr), toning, or denoising.” Yang also discloses noise suppression filtering after decoding: Yang, page 27, paragraph 4 discloses “The loop filter unit 220 may include one or more loop filters, such as a deblocking filter, a sample-adaptive offset (SAO) filter, or one or more other filters, such as self- Adaptive loop filter (ALF), noise suppression filter (NSF), or any combination.” Thus, Yang recites a motivation for incorporating a denoising operation in the process. It would have been obvious to one of ordinary skill in the art to incorporate the denoising in the latent space, with respect to time dependent constants, as taught by Ho. Accordingly, the combination of Yang and Ho discloses the invention of Claim 1.
Regarding Claim 2, the combination of Yang and Ho discloses “The method of claim 1, wherein: the quantized latent representation is entropy coded” (Yang, page 32, paragraph 3 discloses “The entropy coding unit 270 is used for entropy coding algorithm or scheme”), “and the quantized latent representation is entropy decoded before performing the inverse quantization process” (Yang, page 32, paragraph 6 discloses “In the example of FIG. 3, decoder 30 includes entropy decoding unit 304, inverse quantization unit 310”).
Regarding Claim 3, the combination of Yang and Ho discloses “The method of claim 1, further comprising:
receiving a quantization setting that is generated based on the latent representation in the latent space” (Yang, page 54, paragraph 2 discloses “Different quantization step sizes are set, that is, the fidelity values of all coding blocks are quantized using different quantization step sizes. Among them, the quantization step size set by the encoding end needs to be informed to the decoding end; that is, which quantization step size is used for quantization of the fidelity map of the encoded image or which quantization step size is used for quantization of the fidelity value of which encoding block, and the encoding The end needs to inform the decoding end”; where quantization step size is a quantization setting based on the latent representation); “and
performing the inverse quantization process using the quantization setting” (Yang, page 26, paragraph 5 discloses “Quantization may include dividing by the quantization step size, and corresponding or inverse dequantization performed by the inverse quantization unit 210 or the like may include multiplying by the quantization step size.”)
Regarding Claim 4, the combination of Yang and Ho discloses “The method of claim 3, wherein the quantization setting is used by the inverse quantization process to adjust a number of bits that are used for quantization to generate the reconstructed latent representation” (Yang, page 26, paragraph 5 discloses “For example, n-bit transform coefficients may be rounded down to m-bit transform coefficients during quantization, where n is greater than m. The degree of quantization can be modified by adjusting the quantization parameter (QP)” and “The inverse quantization unit 210 is configured to perform inverse quantization of the quantization unit 208 on the quantized coefficients 209 to obtain the dequantized coefficients 211, for example, performing the inverse of the quantization scheme performed by the quantization unit 208 according to or using the same quantization step size as the quantization unit 208”; where quantization step size is a quantization setting).
Regarding Claim 5, the combination of Yang and Ho teaches “The method of claim 3, wherein:
the quantization setting is generated based on the latent representation of the image and an input parameter, wherein the input parameter is based on a rate distortion balance” (Yang, page 54, paragraph 2 discloses “Quantization will bring distortion to the fidelity value. Therefore, when setting the quantization step size at the encoding end, the corresponding quantization step size can be set according to the accuracy requirements of the decoding end for the fidelity of the reconstructed image”).
Regarding Claim 6, the combination of Yang and Ho teaches “The method of claim 1, wherein: the time step parameter is generated based on the latent representation of the image and an input parameter, wherein the input parameter is based on a rate distortion balance” (Yang, page 54, paragraph 2 discloses “Quantization will bring distortion to the fidelity value. Therefore, when setting the quantization step size at the encoding end, the corresponding quantization step size can be set according to the accuracy requirements of the decoding end for the fidelity of the reconstructed image”).
Regarding Claim 7, the combination of Yang and Ho teaches “The method of claim 1, wherein:
a quantization process and inverse quantization process add quantization error to the reconstructed latent representation compared to the latent representation of the image, and the diffusion model is configured to remove noise associated with the quantization error” (Yang, page 40, paragraph 2 discloses “Using the specified quantization step size Qstep to quantize the specified signal, the intensity of the quantization distortion can be obtained by theoretical analysis. For example, assuming a uniformly distributed source, the mean square error of the quantization distortion for uniform scalar quantization is Qstep2 ”; where mean square error of quantization distortion is quantization error added by quantization and inverse quantization processes. Ho, Section 4.3 discloses “Treating the variational bound terms L1+···+LT as rate and L0 as distortion, our CIFAR10 model with the highest quality samples has a rate of 1.78 bits/dim and a distortion of 1.97 bits/dim, which amounts to a root mean squared error of 0.95 on a scale from 0 to 255. More than half of the lossless codelength describes imperceptible distortions”; where Ho discloses reducing distortion).
Regarding Claim 8, the combination of Yang and Ho teaches “The method of claim 1, wherein:
a quantization error from generating the quantized latent representation adds noise to the reconstructed latent representation, and the diffusion model denoises the reconstructed latent representation to remove noise from the reconstructed latent representation” (Yang, page 40, paragraph 2 discloses “Using the specified quantization step size Qstep to quantize the specified signal, the intensity of the quantization distortion can be obtained by theoretical analysis. For example, assuming a uniformly distributed source, the mean square error of the quantization distortion for uniform scalar quantization is Qstep2 ”; where quantization distortion is quantization error. Ho, Section 4.3 discloses “Treating the variational bound terms L1+···+LT as rate and L0 as distortion, our CIFAR10 model with the highest quality samples has a rate of 1.78 bits/dim and a distortion of 1.97 bits/dim, which amounts to a root mean squared error of 0.95 on a scale from 0 to 255. More than half of the lossless codelength describes imperceptible distortions”; where Ho discloses reducing distortion and thus teaches removing noise).
Regarding Claim 12, Claim 12 recites a computer-readable storage medium storing a program with instructions corresponding to the steps recited in Claim 1. Therefore, the recited programming instructions of this claim are mapped to the proposed combination in the same manner as the corresponding steps in its corresponding method claim. Additionally, the rationale and motivation to combine the Yang and Ho references, presented in rejection of Claim 1, apply to this claim. Finally, the combination of Yang and Ho references discloses “A non-transitory computer-readable storage medium having stored thereon computer executable instructions, which when executed by a computing device” (Yang, page 17, paragraph 7 discloses “the application relates to a non-transitory computer-readable storage medium comprising program code, when executed by a computer device, for performing the first, second or third aspect or the first, second or the method in any possible embodiment of the third aspect.”)
Regarding Claim 13, Yang teaches “A method comprising:
receiving an image” (Yang, page 38, paragraph 1 discloses “Figure 6 shows a typical deep neural network-based image encoding and decoding scheme, which adopts an autoencoder network structure. The input x is the original image to be encoded”; where inputting an original image is receiving an image);
“encoding the image into a latent representation in a latent space” (Yang, page 38, paragraph 1 discloses “Figure 6 shows a typical deep neural network-based image encoding and decoding scheme, which adopts an autoencoder network structure. The input x is the original image to be encoded, which can be expressed as an array of wx × hx × cx , where wx , hx , and cx represent the width, height, and number of color components of the input image, respectively. The analyzer performs a dimensionality reduction operation on the input image x to obtain its latent representation y”; )
performing a quantization process on the latent representation to generate a quantized latent representation” (Yang, page 38, paragraph 2 also discloses “The quantization unit 208 in FIG. 2 is configured to perform a quantization operation on the transform coefficient 207 output by the transform processing unit 206 by applying scalar quantization or vector quantization to obtain a quantization level (quantization level) 209 of the transform coefficient 207, wherein the quantization level 209 It is the output of the quantized transform coefficient, also called the quantized transform coefficient 209, or the quantized coefficient 209”); “and
transmitting the quantized latent representation to a receiver, wherein an inverse quantization process is performed to generate a reconstructed latent representation” (Yang, page 27, paragraphs 2-3 disclose “The inverse quantization unit 210 is configured to perform inverse quantization of the quantization unit 208 on the quantized coefficients 209 to obtain the dequantized coefficients 211, for example, performing the inverse of the quantization scheme performed by the quantization unit 208 according to or using the same quantization step size as the quantization unit 208” and “Inverse quantization unit 310 may be operable to receive quantization parameters (QPs) (or information related to inverse quantization in general) and quantization coefficients from encoded image data 21 (eg, parsed and/or decoded by entropy decoding unit 304)” and “The quantization parameters inverse quantize the decoded quantized coefficients 309 to obtain inverse quantized coefficients 311 , which may also be referred to as transform coefficients 311 or dequantized coefficients 311 “; where dequantized coefficients are a reconstructed latent representation) “
Yang does not explicitly teach “estimating a time step parameter based on the latent representation” and “a diffusion model performs a denoising process for a number of iterations based on the time step parameter to remove noise from the reconstructed latent representation.”
However, in an analogous field of endeavor, Ho teaches “estimating a time step parameter based on the latent representation” (Ho Section 2, paragraph 1 discloses
PNG
media_image2.png
195
755
media_image2.png
Greyscale
Ho, Section 3.2 discloses “First, we set Σθ(xt ,t)=σ2tI to untrained time dependent constants”; where untrained time dependent constants are time step parameters generated based on latent representation)
and “a diffusion model performs a denoising process for a number of iterations based on the time step parameter to remove noise from the reconstructed latent representation” (Ho, Section 3.4, paragraph 2 discloses “our diffusion process setup in Section 4 causes the simplified objective to down-weight loss terms corresponding to small t. These terms train the network to denoise data with very small amounts of noise, so it is beneficial to down-weight them so that the network can focus on more difficult denoising tasks at larger t terms”; where denoising data by a diffusion process is performing a denoising process using a diffusion model; see Figure 2, below which shows a diffusion model over iterations 1<t≤T).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Yang to incorporate the teachings of Ho by implementing a diffusion model to perform denoising iteratively. The prior art Yang contained a ‘base’ method upon which the claimed invention can be seen as an ‘improvement.’ Yang teaches an encoder/decoder scheme for video or image compression. The claimed invention recites an encoder/decoder scheme for image compression, and also discloses a diffusion model performing denoising over a number of timesteps before decoding. The prior art contained a known technique that is applicable to the base method. Ho teaches a diffusion model that performs denoising on latent variables: Ho, Section 2 discloses “Diffusion models [53] are latent variable models,” and Ho, Section 3.4, paragraph 2 discloses “our diffusion process setup in Section 4 causes the simplified objective to down-weight loss terms corresponding to small t. These terms train the network to denoise data with very small amounts of noise, so it is beneficial to down-weight them so that the network can focus on more difficult denoising tasks at larger t terms.” One of ordinary skill in the art would have recognized that applying the known technique would have yielded predictable results and resulted in an improved system. That is, it would have been obvious to one of ordinary skill in the art that applying a denoising method designed for latent representations, as taught by Ho, to latent representations of images of Yang would have yielded predictable results of denoising images in the latent space. Additionally, Yang discloses denoising as a preprocessing step: Yang, page 20, paragraph 4 discloses “For example, the preprocessing performed by the preprocessor 18 may include trimming, color format conversion (eg, from RGB to YCbCr), toning, or denoising.” Yang also discloses noise suppression filtering after decoding: Yang, page 27, paragraph 4 discloses “The loop filter unit 220 may include one or more loop filters, such as a deblocking filter, a sample-adaptive offset (SAO) filter, or one or more other filters, such as self- Adaptive loop filter (ALF), noise suppression filter (NSF), or any combination.” Thus, Yang recites a motivation for incorporating a denoising operation in the process. It would have been obvious to one of ordinary skill in the art to incorporate the denoising in the latent space, with respect to time dependent constants, as taught by Ho. Accordingly, the combination of Yang and Ho discloses the invention of Claim 13.
Regarding Claim 14, the combination of Yang and Ho teaches “The method of claim 13, wherein: the quantized latent representation is entropy coded, and the quantized latent representation is entropy decoded before performing the inverse quantization process” (Yang, page 32, paragraph 3 discloses “The entropy coding unit 270 is used for entropy coding algorithm or scheme”), “and the quantized latent representation is entropy decoded before performing the inverse quantization process” (Yang, page 32, paragraph 6 discloses “In the example of FIG. 3, decoder 30 includes entropy decoding unit 304, inverse quantization unit 310”).
Regarding Claim 15, the combination of Yang and Ho discloses “The method of claim 13, further comprising:
determining a quantization setting that is generated based on the latent representation in the latent space” (Yang, page 54, paragraph 2 discloses “Different quantization step sizes are set, that is, the fidelity values of all coding blocks are quantized using different quantization step sizes. Among them, the quantization step size set by the encoding end needs to be informed to the decoding end; that is, which quantization step size is used for quantization of the fidelity map of the encoded image or which quantization step size is used for quantization of the fidelity value of which encoding block, and the encoding The end needs to inform the decoding end”; where quantization step size is a quantization setting based on the latent representation); “and
performing the quantization process using the quantization setting” (Yang, page 26, paragraph 5 discloses “Quantization may include dividing by the quantization step size, and corresponding or inverse dequantization performed by the inverse quantization unit 210 or the like may include multiplying by the quantization step size.”)
Regarding Claim 16, the combination of Yang and Ho teaches “The method of claim 15, wherein:
the quantization setting is generated based on the latent representation of the image and an input parameter, wherein the input parameter is based on a rate distortion tradeoff.” (Yang, page 54, paragraph 2 discloses “Quantization will bring distortion to the fidelity value. Therefore, when setting the quantization step size at the encoding end, the corresponding quantization step size can be set according to the accuracy requirements of the decoding end for the fidelity of the reconstructed image”).
Regarding Claim 17, the combination of Yang and Ho teaches “The method of claim 13, wherein estimating the time step parameter comprises: estimating the time step parameter based on the latent representation of the image and an input parameter, wherein the input parameter is based on a rate distortion tradeoff” (Yang, page 54, paragraph 2 discloses “Quantization will bring distortion to the fidelity value. Therefore, when setting the quantization step size at the encoding end, the corresponding quantization step size can be set according to the accuracy requirements of the decoding end for the fidelity of the reconstructed image”).
Regarding Claim 18, the combination of Yang and Ho teaches “The method of claim 13, wherein:
a quantization error from the quantization process adds noise to the reconstructed latent representation, and the diffusion model denoises the reconstructed latent representation to remove noise from the reconstructed latent representation” (Yang, page 40, paragraph 2 discloses “Using the specified quantization step size Qstep to quantize the specified signal, the intensity of the quantization distortion can be obtained by theoretical analysis. For example, assuming a uniformly distributed source, the mean square error of the quantization distortion for uniform scalar quantization is Qstep2 ”; where quantization distortion is quantization error. Ho, Section 4.3 discloses “Treating the variational bound terms L1+···+LT as rate and L0 as distortion, our CIFAR10 model with the highest quality samples has a rate of 1.78 bits/dim and a distortion of 1.97 bits/dim, which amounts to a root mean squared error of 0.95 on a scale from 0 to 255. More than half of the lossless codelength describes imperceptible distortions”; where Ho discloses reducing distortion and thus teaches removing noise).
Allowable Subject Matter
Claims 9-11, 19, and 20 would be allowable if rewritten to overcome the rejection(s) under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), 2nd paragraph, set forth in this Office action and to include all of the limitations of the base claim and any intervening claims.
Regarding Claim 9, the combination of Yang and Ho does not explicitly teach “The method of claim 1, wherein: a network is trained to generate the time step parameter.”
Ho discloses only “untrained time dependent constants,” and does not explicitly teach a network trained to generate the time step parameter. That is, although the combination of Yang and Ho discloses a diffusion model that operates at particular time steps, none of the cited prior art explicitly teaches a separate network trained to generate “the time step parameter,” as required by Claim 9.
Thus, none of the previously cited prior art, alone or in combination, provides a motivation to teach the ordered combination of “The method of claim 1, wherein: a network is trained to generate the time step parameter.”
Claims 10, 11, and 20 depend from Claim 9 and thus contain all allowable subject matter of Claim 9.
Regarding Claim 19, the combination of Yang and Ho does not explicitly teach “The method of claim 13, wherein: a network is trained to generate the time step parameter.”
Ho discloses only “untrained time dependent constants,” and does not explicitly teach a network trained to generate the time step parameter. That is, although the combination of Yang and Ho discloses a diffusion model that operates at particular time steps, none of the cited prior art explicitly teaches a separate network trained to generate “the time step parameter,” as required by Claim 19.
Thus, none of the previously cited prior art, alone or in combination, provides a motivation to teach the ordered combination of “The method of claim 13, wherein: a network is trained to generate the time step parameter.”
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure: Zhang et al. (CN 110198444 A) discloses a method for video frame encoding including a quantization parameter prediction network and using a time characteristic parameter to obtain the quantization parameter (see Claim 4).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CAROLINE TABANCAY DUFFY whose telephone number is (703)756-1859. The examiner can normally be reached Monday - Friday 8:00 am - 5:30 pm.
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, Amandeep Saini can be reached at 5712723382. 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.
/CAROLINE TABANCAY DUFFY/Examiner, Art Unit 2662
/AMANDEEP SAINI/Supervisory Patent Examiner, Art Unit 2662
Prosecution Insights