Unlocking the Potential of Light for Energy-Efficient Deep Learning

Project Overview: Deep neural networks have undeniably transformed the landscape of artificial intelligence, but their hunger for electronic computing power has always been a challenge. Optical Neural Networks (ONNS), a cutting-edge domain that seeks to capitalize on the immense potential of light’s speed and efficiency. This semester, we invite you to join us in exploring this exciting frontier.

The Challenge: The core challenge in optical neural networks lies in creating multiple optical layers without relying on traditional electronic components and train ONNs.

Our Solution: We present a groundbreaking framework that harnesses multiple scattering to synthesize programmable linear and nonlinear transformations concurrently, all while operating at low optical power. Our approach leverages the unique relationship between the scattering potential (represented by data) and the scattered field. In simpler terms, we’re using the power of light to perform complex computations with higher energy efficiency.

What You’ll Explore:
• Dive deep into the experimental foundations of optical neural networks.
• Gain hands-on experience with experimental investigations that demonstrate the power of ONNs.
• Discover how the repetition of data through multiple scattering enables nonlinear optical computing using continuous wave light, all while consuming minimal energy.

Why Join Us:
• Explore a cutting-edge field at the intersection of optics and deep learning.
• Contribute to the development of energy-efficient AI technologies.

Ready to Illuminate Your Semester? If you’re intrigued by the idea of unleashing the potential of light for energy-efficient deep learning, email Mustafa Yildirim for more details.