Title:  Memrisistive Devices for Neuromorphic Computing Applications

Committee:

Dr. Alan Doolittle, ECE, Chair , Advisor

Dr. Albert Frazier, ECE

Dr. William Hunt, ECE

Dr. Muhannad Bakir, ECE

Dr. Faisal Alamgir, MSE

Abstract:

New computing paradigms are being investigated as the performance of digital computers has begun to saturate due to material, size, and power limitations. Among these new paradigms is neuromorphic computing, which seeks to replicate the computational architecture of the human brain. In order to control the body, process noisy sensory inputs, and recognize patterns within our environment, the human brain performs computations 3 – 6 orders of magnitude faster than a modern processor while only consuming 20W of power. While significant advances have been made replicating the neural architecture using traditional analog electronics, recent advances in memristive devices offer new approaches based on the same ionic physics encountered in the brain.

This thesis seminar will discuss memristive materials and devices that enable biologically realistic behavior without increasing circuit complexity. New physics encountered in ionic memristors will be presented that mimics biological processes such as synaptic plasticity, ionic integration, and opto-ionic coupling. Capitalizing on this physics, memristive devices will be shown that demonstrate short-term and long-term memory, optical detection similar to a retinal cell, and signal adaptation that has been shown to stabilize our brains.