Scalable photonic deep neural networks; from tiled electronic-photonic chips to photonic training (SPDNN)
Abstract
Deep learning using deep neural networks is increasingly in demand for many applications from image and speech recognition to artificial intelligence. The state-of-the-art method for implementation of such networks is to utilize advanced parallel graphics processing units (GPU) where despite excellent performance, it suffers from high power consumption and long processing time for complex multi-layer networks. Here, we propose to work on photonic deep neural networks by conducting 4 tasks; (1) measurement and demonstration of a monolithic photonic deep photonic network with a larger number of neurons for instantaneous classification of high resolutionimages compared, (2) Tape-out and measurements of a highly scalable multi-chip "tiled" photonic deep neural network for instantaneous image classification. (3) design and tape-out of photonic deep networks for instantaneous classification of input digital data, and (4) design and tape-out of photonic back propagation for nanosecond leveltraining of deep neural networks.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 13, 2025
- Source ID
- N000142512126
Entities
People
- Firooz Aflatouni
Organizations
- Office of Naval Research
- United States Navy
- University of Pennsylvania