4_21_09_BayesianInference

# 4_21_09_BayesianInference - Computations performed by...

This preview shows pages 1–7. Sign up to view the full content.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Computations performed by neural networks can be expressed as energy minimization.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
A link exists between energy minimization and Bayesian processes (and therefore, between these things and neural networks). P V ( ) = e −β E V ( ) e −β E V ( ) V If a system can be in a finite number of states, the transition between them tends to minimize energy (E( V )) except for noise, and the transition probabilities depend only on the current state, then in steady state The Gibbs distribution
Neural Response ( r ) Sensory Input (s) a b c Bayes Theorem P s | r ( ) = P ( r | s ) P s ( ) P r ( ) b a + b = b b + c b + c a + b + c a + b a + b + c P r ( ) = a + b a + b + c P s ( ) = b + c a + b + c P r | s ( ) = b b + c b a + b P s | r ( ) =

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Response ( r ) Sensory Input (s) a b c Bayes Theorem Hence, one can often replace a network computation by a Bayesian one, thus understand- ing the computation and the noise. P s
This is the end of the preview. Sign up to access the rest of the document.

## This note was uploaded on 06/08/2009 for the course BME 575L taught by Professor Grzywacz during the Spring '09 term at USC.

### Page1 / 26

4_21_09_BayesianInference - Computations performed by...

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online