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 .
Priority
Applicant claims the benefit of US Provisional Application No. 63/589,248, filed 10/10/2023. The effective filing date of 10/10/2023 has been used for claims 1-9. Claims 10-12 contain subject matter not disclosed in the provisional application, thus the effective filing date of 10/04/2024 has been used for claims 10-12.
If applicant intends to have these claims afforded the benefit of the earlier filing date, applicant may: (1) amend the claim limitation(s) to avoid it/them containing subject matter not in the provisional application; or (2) present a sufficient showing that the claim limitation(s) recite(s) material from specific sections of the provisional application.
Specification
The disclosure is objected to because of the following informalities:
In para 59, “you can then then transmit pre-frame” should read “you can then
In para 129: “a half-precision VAE fix was used utilized” should read “a half-precision VAE fix was used .
Appropriate correction is required.
The lengthy specification has not been checked to the extent necessary to determine the presence of all possible minor errors. Applicant’s cooperation is requested in correcting any errors of which applicant may become aware in the specification.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1, 3, 5, and 7 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Chen et al. (Chen, T., Zhang, R., & Hinton, G. (2022). Analog bits: Generating discrete data using diffusion models with self-conditioning. arXiv preprint arXiv:2208.04202.), hereinafter Chen.
Regarding claim 1, Chen teaches a computer-implemented method (Chen, pg. 4-5, section 3), comprising: receiving an input image including a plurality of pixels where each of the plurality of pixels is represented by multiple bits (Chen, each pixel represented as a token, see Binary encoding in section 3.1, pg. 4, section 3.1: “Each pixel consists of 3 sub-pixels (RGB channels), and each sub-pixel is an integer in [0,256) representing the intensity…In UINT8, we use 8-bit binary codes converted from the corresponding sub-pixel integer in [0,256)”);
transforming the multiple bits representing each of the plurality of pixels of the input image into a set of floating-point values (Chen, generation of analog bits from binary codes, see FIG. 1 attached below; abstract: “The main idea behind our approach is to first represent the discrete data as binary bits, and then train a continuous diffusion model to model these bits as real numbers which we call analog bits”; see last para on pg. 2);
providing the set of floating-point values for each of the plurality of pixels of the input image to a denoising model of a machine-trained diffusion model (Chen, see input to diffusion model in FIG. 1 and “denoising” and “diffusion models” in pg. 2, section 2);
generating, by the denoising model, successive sets of floating-point values (Chen, see generated output in FIG. 1; pg. 2, 1st para: “generated analog bits”; successive floating-point values generated by generating output analog bits for each token/pixel – application of FIG. 1 model to each token, abstract: “IMAGENET 64×64 (which has 12K discrete 8-bit tokens)”); and
reconstructing the plurality of pixels of the input image from the successive sets of floating-point values (Chen, pg. 2, 1st para: “At sampling time, the generated analog bits can be decoded into discrete variables by a simple thresholding operation…With strong continuous generative models (diffusion models in particular), it should not be too difficult to generate highly concentrated bimodal data where each real-valued analog bit is close to a binary bit”; see Bit Diffusion image generation in FIG. 4, pg. 6, 2nd para: “Figure 4 shows generated images of different diffusion models on continuous and discrete IMAGENET 64×64”).
PNG
media_image1.png
173
825
media_image1.png
Greyscale
Regarding claim 3 (dependent on claim 1), Chen teaches wherein the reconstructing further includes converting the successive sets of floating-point values generated by the denoising model into successive sets of binary values wherein each of the successive sets of floating-point values corresponds to one of the plurality of pixels of the input image (Chen, output set of generated analog bits corresponds to the input token pixel; all floating-point values are converted to binary values via thresholding, pg. 2, 1st para: “At sampling time, the generated analog bits can be decoded into discrete variables by a simple thresholding operation”).
Regarding claim 5, Chen teaches a computing system (Chen, architecture in pg. 4-5, section 3), comprising: one or more processors; and one or more non-transitory, computer-readable media storing a machine-implemented diffusion model including a denoising model and instructions that, when executed by the one or more processors, cause the one or more processors to perform the claimed steps (Chen, computing components utilized to train the diffusion models and execute image generation, see “denoising” and “diffusion models” in pg. 2, section 2; see stored algorithms on pg. 4 and architecture on pg. 5). All further claim limitations executed by the processor(s) are met by Chen because the executed steps of claim 5 are the same as the method steps in claim 1.
Regarding claim 7 (dependent on claim 5), all claim limitations are met by Chen because the limitations of claim 7 are the same as the method steps of claim 3.
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.
Claims 2, 4, 6, and 8-11 are rejected under 35 U.S.C. 103 as being unpatentable over Chen in view of Moyer (Moyer, Bryon. Addressing a Single Bit, 18 Sept 2020, INSIDE THE IOT [online], [retrieved on 2026-07-15]. Retrieved from the Internet <URL: https://www.insidetheiot.com/addressing-a-single-bit/>).
Regarding claim 2 (dependent on claim 1), Chen teaches wherein the transforming includes, for each pixel of the plurality of pixels of the input image: converting integer outputs into the set of floating-point values for the pixel (Chen, conversion of bit integers to analog bits in FIG. 1, pg. 2, last para: “we propose to simply cast the binary bits {0,1}n into real numbers Rn for the continuous diffusion models 1. We term these real numbers analog bits since they learn to share the same bimodal values as binary bits but are modeled as real numbers”).
Chen teaches bits arranged in parallel when input and processed by diffusion models (Chen, see FIG. 1 input of bits to “diffusion models”); however, Chen fails to explicitly teach applying multiple bit masks arranged in parallel to the multiple bits of the pixel wherein different ones of the bits masks are applied to different ones of the multiple bits of the pixel; converting integer outputs resulting from the applying of the multiple bit masks into the set of floating-point values for the pixel.
However, Moyer teaches a method for applying multiple bit masks to multiple bits wherein different ones of the bits masks are applied to different ones of the multiple bits (Moyer, pg. 2-3, sections “Detecting a 0 Bit” and “Detecting a 1 Bit” – see attachments below where different masks are applied to different bit positions).
PNG
media_image2.png
623
655
media_image2.png
Greyscale
Applying bit masks allows the model to select/ignore certain bits in a set of binary values. Chen discloses a base method for processing multiple bits in parallel, but does not specify specific methods for bit-masking to isolate certain values. Moyer teaches a known technique of bit-masking to obtain a single bit value at a specific position. A person having ordinary skill in the art, before the effective filing date of the claimed invention, could have applied the known technique, as taught by Moyer, in the same way to the method of Chen and achieved predictable results of isolating single bit values before conversion to floating-point values and input to the diffusion model.
Regarding claim 4 (dependent on claim 3), Chen teaches wherein the reconstructing further includes, for each successive set of binary values: generate multiple reconstructed bits of one pixel of the plurality of pixels of the input image (Chen, bit2int operation, see FIG.1 and the algorithms at the top of pg. 4), but fails to explicitly teach multiplying each binary value of each successive set of binary values by a different one of multiple bit masks.
However, Moyer teaches multiplying each binary value of a set of binary values by a different one of multiple bit masks (Moyer, pg. 2-3, sections “Detecting a 0 Bit” and “Detecting a 1 Bit” – see attachment in claim 3 where different masks are applied to different bit positions).
Applying bit masks allows the model to select/ignore certain bits in a set of binary values. Chen discloses a base method for processing multiple bits in parallel, but does not specify specific methods for bit-masking to isolate certain values. Moyer teaches a known technique of bit-masking to obtain a single bit value at a specific position. A person having ordinary skill in the art, before the effective filing date of the claimed invention, could have applied the known technique, as taught by Moyer, in the same way to the method of Chen and achieved predictable results of isolating single bit values for subsequent reconstruction of the final pixel value.
In the method taught by Chen, adding the results of multiplying each binary value by a different bit mask produces the same output of discrete values obtained after thresholding. These values are then reconstructed using the bit2int operation.
Regarding claim 6 (dependent on claim 5), all claim limitations are met and rendered obvious by Chen in view of Moyer because the limitations of claim 6 are the same as the method steps of claim 2.
Regarding claim 8 (dependent on claim 7), all claim limitations are met and rendered obvious by Chen in view of Moyer because the limitations of claim 8 are the same as the method steps of claim 4.
Regarding claim 9 (dependent on claim 8), Chen in view of Moyer teaches further comprising training or fine tuning with training imagery to optimize performance (Chen, training of Bit Diffusion model described at top of pg. 4; see also use of IMAGENET imagery in section 3.1 on pg. 4).
Regarding claim 10 (dependent on claim 8) Chen in view of Moyer teaches further comprising re-weighting different connections of the denoising model per parallel branch (Chen, Chen teaches, exemplified in FIG. 1, input to a plurality of diffusion models, which may be considered parallel branches; see weights during training in section 3.1 on pg. 5).
Regarding claim 11 (dependent on claim 8) Chen in view of Moyer teaches further comprising employing different masking and quantization strategies (Moyer, different masking strategies – masking more than just one bit at a time, top of pg. 3: “you can mask to see more than just one bit also.”).
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Chen in view of Moyer, in further view of Gonzalez et al. (Gonzalez, R. C. and Woods, R. E. Digital Image Processing (New York, 2018), Pearson, pp. 203-213.), hereinafter Gonzalez.
Regarding claim 12 (dependent on claim 8) Chen in view of Moyer fails to explicitly teach further comprising transforming input data to another domain, including to the frequency domain via a Fourier transform variant, to form an output; and inverse transforming the output.
However, Gonzalez teaches an image processing method of transforming input data to another domain, including to the frequency domain via a Fourier transform variant, to form an output; and inverse transforming the output (Gonzalez, transforming an image to the frequency domain to perform filtering, pg. 204, 3rd para: “Both representations share the important characteristic that a function, expressed in either a Fourier series or transform, can be reconstructed (recovered) completely via an inverse process, with no loss of information. This is one of the most important characteristics of these representations because it allows us to work in the Fourier domain (generally called the frequency domain) and then return to the original domain of the function without losing any information.”).
Chen discloses a base method for inputting image data into a machine-learned model, but does not specify specific methods for image pre- or post-processing. Gonzalez teaches a known technique of filtering an image in the frequency domain to enhance the image, for example, via sharpening and/or smoothing techniques. A person having ordinary skill in the art, before the effective filing date of the claimed invention, could have applied the known technique, as taught by Gonzalez, in the same way to the method of Chen in view of Moyer and achieved predictable results of enhancing a captured image before input to the machine-learned model in order to improve the image processing results of the model.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EMMA E DRYDEN whose telephone number is (571)272-1179. The examiner can normally be reached M-F 9-5 EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, ANDREW BEE can be reached at (571) 270-5183. 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.
/EMMA E DRYDEN/Examiner, Art Unit 2677
/ANDREW W BEE/Supervisory Patent Examiner, Art Unit 2677
Prosecution Insights