| Description |
-
Thesis (M.S.)--Tufts
University, 2020.
Submitted to the Dept. of
Electrical Engineering.
Advisor: Sameer
Sonkusale.
Committee: Jeffery Hopwood, and Gouqing
Fu.
Keyword: Electrical
engineering.
In this work I present an energy
efficient mixed-signal realization of a quantized neural network. In this implementation
I represent signals in the time ... read moredomain. By encoding signals as pulse-widths, we can take
advantage of the extremely accurate clocks and delay resolution in nanometer CMOS
processes. Weights are represented as binary weighted current sources with local DRAM
storage. Capacitive integration implements weighted addition, and an inverter provides
as rectified linear activation. I present the design in 0.18um CMOS process with
successful circuit and behavioral simulation results on the IRIS
dataset.read less
|
|---|
This object is in collection
