UCLA researchers have conducted an in-depth analysis of nonlinear information coding strategies for diffractive optical processors, providing new insights into their performance and utility. Their study, published in Light: Science & Applications, compared simpler nonlinear coding strategies, such as phase coding, with the performance of nonlinear information coding methods based on data repetition, highlighting their advantages and limitations in optical processing of visual information.
Diffractive optical processors, built from linear materials, perform computational tasks by manipulating light using structured surfaces. Nonlinear encoding of optical information can improve the performance of these processors, allowing them to better handle complex tasks such as image classification, quantitative phase imaging, and encryption.
The UCLA research team, led by Professor Aydogan Ozcan, evaluated various nonlinear coding strategies using different datasets to assess their statistical inference performance. Their results revealed that data repetition within a diffractive volume, while improving inference accuracy, compromises the universal linear transformation capability of diffractive optical processors. Therefore, data repetition-based diffractive blocks cannot serve as optical analogues to the fully connected or convolutional layers commonly used in digital neural networks. More generally, data repetition-based diffractive processors can be viewed as a simplified optical analogue of the dynamic convolutional kernel concept used in some neural network architectures. Despite its different characteristics, the data repetition architecture within a diffractive optical processor remains effective for inference tasks and offers advantages in terms of noise resilience.
Alternatively, phase encoding of input information without data repetition offers a simpler nonlinear encoding strategy to implement with statistically comparable inference accuracy. Implemented directly via spatial light modulators or phase-only objects, phase encoding is a practical alternative due to its simplicity and efficiency. Additionally, diffractive processors without data repetition do not require preprocessing of the input information via a digital system, which is necessary for visual data repetition. Therefore, data repetition can be time-consuming, especially for phase-only input objects, due to the need for digital phase recovery and preprocessing before visual data repetition can occur.
The research team’s results provide valuable insights into the push-pull relationship between linear material-based diffractive optical systems and nonlinear information coding strategies. These results offer potential for a wide range of applications, including optical communications, surveillance, and computational imaging. The ability to improve inference accuracy through nonlinear coding strategies can improve the performance of optical processors in various fields, leading to more advanced and efficient visual information processing systems.
The authors of this paper are Yuhang Li, Jingxi Li, and Aydogan Ozcan, all affiliated with the UCLA Department of Electrical and Computer Engineering. Professor Ozcan is also an associate director of the California NanoSystems Institute (CNSI).
This research was funded by the DOE (United States).
/Public dissemination. This content from the original organization/authors may be of a timely nature and edited for clarity, style, and length. Mirage.News takes no institutional position or bias, and all views, positions, and conclusions expressed herein are solely those of the author(s). See the full story here.