Computers - Using your Brains. Jim Austin Professor of Neural Computation. Pentium III. So how complex is it ? 10 12 neurons … 1,000,000,000,000 1000 connections between neurons. One brain can hold ... 1,000,000,000,000,000 numbers !. What do 10 12 neurons look like ?.
Professor of Neural Computation
Boolean Logic - both inputs OK, output not OK
=Gates - NAND
ALL inputs to be OK for output to be NOT OK
Should have picked a NAND gate for the brain...
Output = threshold (input A x weight A + input B x weight B)
Threshold logic - threshold 1 - one or more inputs OK output OK
At least one OK for output to be OK
At least three OK’s for output to be OK
using the weights on the inputsWeights
k inputsFunctional elements.
Threshold n gate
Learning at neuron level =
Adjustment of which inputs are important
Conventional computers have no implicit learning ability
Threshold = 2
Better tolerance to failure
Parallelism/use of threshold logic/distributed memory
Better access to uncertain information
Where the inputs are uncertain
Where we want low power
Use of weights and learning methods.
28 Processor cards, each holding 128 hardware neurons.
Each with 1,000,000,000 weights.
PCI based card.
400,000,000 neuron evaluations per second
30 bits set on input
5,120,000,000 neuron weights, 640 neurons.
and many others ….
(It’s Brains from Thunderbirds !)