Reducing Accuracy Loss from Radiation in Analog Memristive Chips for Image Recognition

Electrical Engineering

Team 38: Hoyeol Bae, Wataru Tamaki, Benjamin Tung, Justin Weidmann, Zach White

Summary

This project aims to use analog memristive chips to perform more efficient image recognition. However, conductance drift due to radiation greatly decreases its accuracy. By using techniques to compensate for drift and noise, we are able to improve its accuracy drastically for applications in space where radiation can be a problem.

Video

Research poster

Sponsor

Advisor

Portrait of Matthew Marinella

Matthew Marinella

Associate Professor

School of Electrical, Computer and Energy Engineering

[email protected]