CSC321: Neural Networks
 

Lecture 2: Two simple learning algorithms

Supervised Learning

Linear neurons

A motivating example

Two ways to solve the equations

The cashier’s brain

A model of the cashier’s brain
with arbitrary initial weights

Behaviour of the iterative learning procedure

Deriving the delta rule

The error surface

Online versus batch learning

Adding biases

Binary threshold neurons

The perceptron convergence procedure: Training binary output neurons as classifiers

Weight space

Why the learning procedure works

What binary threshold neurons cannot do