Local Learning in Silicon Integrated Circuits
A talk presented on 2/20/98
to the Robotics & Controls Colloquium
at The University of Washington
Click here to start
Table of Contents
Local Learning in Silicon Integrated Circuits
Neuronal computation
Neuronal computation is not digital
If we could only build machines like this...
We can learn from biology
Neuromorphic engineering
Local learning in neurobiology
A connectionist model
Floating-gate MOSFETs
Electron tunneling
Electron injection
A synapse transistor
Floating-gate transistors multiply
Oxide currents are exponentials
Synapse weight updates are power law
A learning array
Constrained local silicon learning
The learning rule
Silicon learning emulates neurobiology
A model of the primate oculomotor system
The next steps
Author:
Chris Diorio
Email:
diorio@cs.washington.edu
Home Page:
http://www.cs.washington.edu/people/faculty/diorio/