Task-specific sensor optical designs
| DWPI Title: Method for designing compressive sensing matrix for machine learning for performing image classification or regression used in fields e.g. self-driving cars, involves applying machine learning classification algorithm to compressed representation of image |
| Abstract: A method and system architecture for designing a compressive sensing matrix for machine learning includes receiving an image associated with a classification task and; generating a sensing matrix. The sensing matrix includes an array of nonzero elements of the image. A prism array of prism elements is in communication with the sensing matrix. A row of values corresponding with an input angle of the prism array is mapped to a respective column corresponding with a detector. Then the detector detects light refracted at an output angle dictated by the physical shape of the prism element. A physical model of the detector is fabricated and generates a compressed representation of the image. A machine learning classification algorithm is applied to the compressed representation of the image and generates an optimized non-invertible final determination of the image. |
| Use: Method for designing a compressive sensing matrix for machine learning for performing image classification or regression used in fields e.g. self-driving cars, facial recognition, medical imaging and remote sensing. |
| Advantage: The method enables maintaining high performance for a task and object detection or classification, performing classification of the images using machine learning techniques, maintaining dimensionality of the data, recording the smaller subset of data without loss of performance and providing optimal design of optical architectures to realize optimized compressive sensing matrices. The method enables realizing improved performance without increasing the number of prism elements, improving classification accuracy and minimizing classification error. |
| Novelty: The method (200) involves receiving an image associated with a classification task. A sensing matrix is generated (210), where the sensing matrix comprises an array of nonzero elements of the image. A prism array comprising prism elements is provided. A row of values corresponding with an input angle of the prism array is mapped to a respective column corresponding with a detector. An output angle of a prism element of the prism array associated with the input angle is detected by a detector. A physical model of the detector is fabricated. A compressed representation of the image is generated by the physical model. A machine learning classification algorithm is applied to the compressed representation of the image. An optimized non-invertible final determination rate of the image is generated. |
| Filed: 4/4/2022 |
| Application Number: US17712316A |
| Tech ID: SD 15019.0 |
| This invention was made with Government support under Contract No. DE-NA0003525 awarded by the United States Department of Energy/National Nuclear Security Administration. The Government has certain rights in the invention. |
| Data from Derwent World Patents Index, provided by Clarivate All rights reserved. Republication or redistribution of Clarivate content, including by framing or similar means, is prohibited without the prior written consent of Clarivate. Clarivate and its logo, as well as all other trademarks used herein are trademarks of their respective owners and used under license. |