arXiv subjects

Sai Nikhilesh Murty Kottapalli

Publications and source records attributed to Sai Nikhilesh Murty Kottapalli.

3 recordsLinked to original sources

arXiv2026

Multi-channel free-space optical convolutions with incoherent light

Free-space optical systems are promising candidates for high performance computing and have been particularly successful in the implementation of large-scale convolutions. Convolutions are the key operation in convolutional layers, which are used extensively in modern neural networks, especially in the context of image/video processing and generation. These optical accelerators have demonstrated remarkable performance in both processing rates and energy efficiency. Prior approaches have primarily demonstrated convolutions from a single input channel to one or more output channels. We extend these methods to perform true multi-channel convolutions, where multiple input channels are convolved with their own sets of convolutional kernels onto output channels. We simulate this approach using both ray-tracing and angular spectrum propagation and find the approach is highly-scalable. We then experimentally implement a proof-of-concept prototype to demonstrate multi-channel free-space optical convolutions.

physics.optics↗

arXiv2025

Engineering wavefronts with machine learned structured polarization

Optical approaches for wavefront shaping traditionally rely on phase modulation through holographic techniques. Shaping the phase determines a wave's diffraction and hence its intensity distribution in space. We instead show that shaping the polarization introduces a novel framework that permits the spatial modulation of polarization to control wavefront propagation and resulting amplitude distributions. We develop two distinct computational phase retrieval approaches for calculating the required polarization transformations and experimentally validate these. The first method extends the established Gerchberg-Saxton algorithm, while the second employs machine learning optimization to determine optimal polarization patterns. By implementing both amplitude and polarization control simultaneously using a single polarization mask, our approach significantly reduces system complexity compared to traditional methods. Our experimental results demonstrate the potential of polarization-based wavefront shaping as an efficient alternative to conventional techniques, paving the way for applications in optical manipulation and imaging.

physics.optics↗

arXiv2024

Low-power scalable multilayer optoelectronic neural networks enabled with incoherent light

Optical approaches have made great strides towards the goal of high-speed, energy-efficient computing necessary for modern deep learning and AI applications. Read-in and read-out of data, however, limit the overall performance of existing approaches. This study introduces a multilayer optoelectronic computing framework that alternates between optical and optoelectronic layers to implement matrix-vector multiplications and rectified linear functions, respectively. Our framework is designed for real-time, parallelized operations, leveraging 2D arrays of LEDs and photodetectors connected via independent analog electronics. We experimentally demonstrate this approach using a system with a three-layer network with two hidden layers and operate it to recognize images from the MNIST database with a recognition accuracy of 92% and classify classes from a nonlinear spiral data with 86% accuracy. By implementing multiple layers of a deep neural network simultaneously, our approach significantly reduces the number of read-ins and read-outs required and paves the way for scalable optical accelerators requiring ultra low energy.

cs.ET↗