Dissertation Talk: Unconventional Low-Power Computing
Seminar | May 3 | 2-3 p.m. | 540AB Cory Hall
In this talk, I discuss how we can exploit properties that emerge from the collective behavior of constituents at different length scales to reduce power consumption below conventional limits in computing. At the device level, I show how ferroelectric-dielectric coupling ("negative capacitance") can reduce energy consumption below 1/2 CV^2 in capacitors. At the circuit level, high-dimensional distributed representations relax requirements on signal-to-noise ratio and supply voltage, and enable new architecture designs. I describe operations for computing with these representations ("hyperdimensional computing") and show that they are natural for performing energy efficient cognitive computing at the application level.