IP Library › Granted Patent US 12,299,854
Granted Patent B2
US 12,299,854 · App. 18/046,061 · Granted May 13, 2025

Reliability for machine-learning based image generation

Inventors: Alexandru Ardel (Austin, TX); Elad Liebman (Austin, TX); Mrinal Sen (Austin, TX); Georgios Alexandros Dimakis (Austin, TX); Yash Gandhi (Austin, TX); Sriram Ravula (Plano, TX); Dimitri Voytan (Austin, TX)
Assignee: SPARKCOGNITION, INC.
G06T5/70G06N20/00G06T5/00G06T5/50G06T5/77G06T11/006G06V10/25G06V10/751G06V10/758G06T2200/24G06T2207/20081G06T2207/20216
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,299,854
App. No.
18/046,061
Granted
May 13, 2025
Kind
B2
Abstract

A method includes using a machine-learning model to determine multiple sets of image data, each representing an estimated solution to an inverse problem associated with multiple waveform return measurements. First image data are based on a first set of waveform return measurements and first model parameters of the machine-learning model, and second image data are based on a second set of waveform return measurements and a second model parameters of the machine-learning model. The method also includes determining, based on the multiple sets of image data, a representative image. The method further includes generating output data that identifies a first area of the representative image as less reliable than a second area of the representative image based on a statistical evaluation of two or more sets of image data of the multiple sets of image data.

Claims (48)

1. A method comprising:

determining, by one or more processors using a machine-learning model, multiple sets of image data, wherein each set of image data of the multiple sets of image data represents an estimated solution to an inverse problem associated with multiple waveform return measurements, and wherein determining the multiple sets of image data comprises:

determining first image data of the multiple sets of image data based on a first set of waveform return measurements and using a first set of model parameters of the machine-learning model; and

determining second image data of the multiple sets of image data based on a second set of waveform return measurements and using a second set of model parameters of the machine-learning model;

determining, by the one or more processors based on the multiple sets of image data, a representative image; and

generating, by the one or more processors, output data that identifies a first area of the representative image as less reliable than a second area of the representative image based on a statistical evaluation of two or more sets of image data of the multiple sets of image data.

2. The method of claim 1 , wherein the multiple sets of image data include the first image data, the second image data, and one or more additional sets of image data.

3. The method of claim 1 , wherein the representative image corresponds to an average image based on the multiple sets of image data.

4. The method of claim 1 , wherein generating the output data comprises:

determining statistics representing region-by-region variation across the multiple sets of image data; and

identifying regions associated with higher variability as less reliable than regions associated with lower variability.

5. The method of claim 1 , wherein generating the output data comprises:

determining, based on the multiple sets of image data, pixel-by-pixel variability characteristics of the representative image; and

assigning groups of pixels to the first area or the second area based on the pixel-by-pixel variability characteristics.

6. The method of claim 1 , wherein the output data includes a graphical user interface depicting the representative image and visually distinguishing the first area from the second area.

7. The method of claim 1 , wherein the first set of waveform return measurements is distinct from the second set of waveform return measurements.

8. The method of claim 1 , wherein the first set of model parameters is distinct from the second set of model parameters.

9. The method of claim 1 , wherein determining the first image data comprises performing multiple iterations of a gradient descent artifact reduction process, wherein a particular iteration of the gradient descent artifact reduction process includes:

determining, using the machine-learning model, a gradient associated with a particular estimated solution to the inverse problem; and

adjusting the particular estimated solution to the inverse problem based on the gradient to generate an updated estimated solution.

10. The method of claim 9 , wherein the multiple iterations of the gradient descent artifact reduction process are initialized based on solution data determined via reverse time migration based on at least a subset of the multiple waveform return measurements.

11. The method of claim 1 , wherein each image of the multiple sets of image data corresponds to a reflectivity image based on at least a subset of the waveform return measurements.

12. The method of claim 1 , wherein the machine-learning model corresponds to a score-matching network.

13. A system comprising:

one or more processors configured to:

determine, using a machine-learning model, multiple sets of image data, wherein each set of image data of the multiple sets of image data represents an estimated solution to an inverse problem associated with multiple waveform return measurements, and wherein determination of the multiple sets of image data comprises:

determination of first image data of the multiple sets of image data based on a first set of waveform return measurements and using a first set of model parameters of the machine-learning model; and

determination of second image data of the multiple sets of image data based on a second set of waveform return measurements and using a second set of model parameters of the machine-learning model;

determine, based on the multiple sets of image data, a representative image; and

generate output data that identifies a first area of the representative image as less reliable than a second area of the representative image based on a statistical evaluation of two or more sets of image data of the multiple sets of image data.

14. The system of claim 13 , wherein the multiple sets of image data include the first image data, the second image data, and one or more additional sets of image data.

15. The system of claim 13 , wherein the representative image corresponds to an average image based on the multiple sets of image data.

16. The system of claim 13 , wherein to generate the output data the one or more processors are configured to:

determine statistics representing region-by-region variation across the multiple sets of image data; and

identify regions associated with higher variability as less reliable than regions associated with lower variability.

17. The system of claim 13 , wherein to generate the output data the one or more processors are configured to:

determine, based on the multiple sets of image data, pixel-by-pixel variability characteristics of the representative image; and

assign groups of pixels to the first area or the second area based on the pixel-by-pixel variability characteristics.

18. The system of claim 13 , wherein the output data includes a graphical user interface depicting the representative image and visually distinguishing the first area from the second area.

19. The system of claim 13 , wherein to determine the first image data the one or more processors are configured to perform multiple iterations of a gradient descent artifact reduction process, wherein a particular iteration of the gradient descent artifact reduction process includes:

determining, using the machine-learning model, a gradient associated with a particular estimated solution to the inverse problem; and

adjusting the particular estimated solution to the inverse problem based on the gradient to generate an updated estimated solution.

20. A computer-readable storage device storing instructions that, when executed by one or more processors, cause the one or more processors to:

determine, using a machine-learning model, multiple sets of image data, wherein each set of image data of the multiple sets of image data represents an estimated solution to an inverse problem associated with multiple waveform return measurements, and wherein determining the multiple sets of image data comprises:

determination of first image data of the multiple sets of image data based on a first set of waveform return measurements and using a first set of model parameters of the machine-learning model; and

determination of second image data of the multiple sets of image data based on a second set of waveform return measurements and using a second set of model parameters of the machine-learning model;

determine, based on the multiple sets of image data, a representative image; and

generate output data that identifies a first area of the representative image as less reliable than a second area of the representative image based on a statistical evaluation of two or more sets of image data of the multiple sets of image data.

Assignments (2)
CHANGE OF NAME Recorded Jul 17, 2025
From: SPARKCOGNITION, INC.
To: AVATHON, INC.
Reel/Frame 072016/0432 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jan 19, 2023
From: ARDEL, ALEXANDRU; LIEBMAN, ELAD; SEN, MRINAL K; DIMAKIS, GEORGIOS ALEXANDROS; RAVULA, SRIRAM; GANDHI, YASH; VOYTAN, DIMITRI
To: SPARKCOGNITION, INC.
Reel/Frame 062425/0500 →
Continuity (6)
Provisional Application 63362789 · Apr 11, 2022
Provisional Application 63362792 · Apr 11, 2022
Provisional Application 63362787 · Apr 11, 2022
Provisional Application 63263532 · Nov 4, 2021
Provisional Application 63255322 · Oct 13, 2021
Related Publication 20230109854A1 · Apr 13, 2023
References Cited (20)
US 20130126759A1 · Betzig · 2013 [cited by examiner]
CN 116663173A · 2023 [cited by examiner]
Yonel, B., Mason, E., & Yazici, B. (2019). Deep learning for waveform estimation and imaging in passive radar. IET Radar, Sonar & Navigation, 13(6), 915-926. (Year: 2019). [cited by examiner]
Bansal, Arpit et al., “Cold Diffusion: Inverting Arbitrary Image Transforms Without Noise,” Cornell University, Aug. 19, 2022, arXiv:2208.09392v1 [cs.CV], pp. 1-22. [cited by applicant]
Bashir, Y. et al., “Enhancements in Seismic Imaging using Diffraction Studies and Hybrid Traveltime Techniques for PSDM,”IOP Conf. Series: Earth and Environmental Science vol. 38, 2016, AeorEarth, doi:10.1088/1755-1315/… [cited by applicant]
Bora, Ashish et al., “Compressed Sensing using Generative Models,” arXiv:1703.03208v1 [stat.ML] Mar. 9, 2017, https://www.cs.utexas.edu/˜ecprice/papers/compressed-generative.pdf, pp. 1-24. [cited by applicant]
Chen, Chun-Fu (Richard), et al, “CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classification,” https://arxiv.org/abs/2103.14899v2, Aug. 22, 2021, pp. 1-12. [cited by applicant]
Daras, Giannis et al, “Intermediate Layer Optimization for Inverse Problems using Deep Generative Models,” arXiv:2102.07364v1 [cs.LG] Feb. 15, 2021, http://proceedings.mlr.press/v139/daras21a/daras21a.pdf, pp. 1-21. [cited by applicant]
Felzenszwalb, Pedro F. et al., “Efficient Graph-Based Image Sementation,” retrieved Dec. 30, 2021, http://people.cs.uchicago.edu/˜pff/papers/seg-ijcv.pdf, pp. 1-26. [cited by applicant]
Harvey, James E. et al., “Modeling physical optics phenomena by complex ray tracing,” Optical Engineering vol. 54, No. 3, Mar. 2015, pp. 035105-1-0351505-12. [cited by applicant]
He, Kaiming et al., “Deep Residual Learning for Image Recognition,” arXiv:1512.03385v1 [cs.CV] Dec. 10, 2015, https://www.semanticscholar.org/reader/2c03df8b48bf3fa39054345bafabfeff15bfd11d, pp. 1-12. [cited by applicant]
Kaur, Harprett et al., “Improving the resolution of migrated images by approximating the inverse Hessian using deep learning,” Geophysics vol. 85, No. 4 (Jul.-Aug. 2020), pp. WA173-WA183. [cited by applicant]
Menon, Sachit et al., “PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models,” arXiv:2003.03808v3 [cs.CV] Jul. 20, 2020, p. 1-20. [cited by applicant]
Nvidia, Tero Karras et al., “Analyzing and Improving the Image Quality of StyleGAN,” arXiv:1912.04958v2 [cs. CV] Mar. 23, 2020, pp. 1-21. [cited by applicant]
Song, Yang et al., “Generative Modeling by Estimating Gradients of the Data Distribution,” 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), Vancouver, Canada, arXiv:1907.05600v2 [cs.LG] Oct. 29, … [cited by applicant]
Song, Yang et al., “Score-Based Generative Modeling Through Stochastic Differential Equations,” Published as a Conference Paper at ICLR 2021, arXiv:2011.13456v2 [cs.LG] Feb. 10, 2021, pp. 1-36. [cited by applicant]
Tanushev, Nicolay M. et al., “Gaussian beam decompostion for seismic migration,” SEG Technical Program Expanded Abstracts, ResearchGate, Jan. 2011, pp. 1-6. [cited by applicant]
Ulyanov, Dimitry et al., “Deep Image Prior,” arXiv:1711.10925v4 [cs.CV] May 17, 2020, https://dmitryulyanov.github.io/deep_image_prior, pp. 1-23. [cited by applicant]
Vamaraju, Janaki et al., “Accelerating Least Squares Imaging Using Deep Learning Techniques,” Second Workshop on Machine Learning and the Physical Sciences (NeurIPS 2019), Vancouver, Canada. [cited by applicant]
Zeng, Chong et al., “A guide to least-squares reverse time migration for subsalt imaging: Challenges and solutions,” Interpretation, Aug. 2017, pp. SN1-SN11. [cited by applicant]