Perceptron learning

• Find weight values by iterative improvement

  1. Cycle through the input patterns
  
  2. Is the output correct?
  ?Yes: Go to the next pattern
  
  ? No: Modify the connection weights using a Hebbian-like rule

• Learning rule converges in a finite number of steps

  • If a solution exist

• More about this next week

Previous slide

Next slide

Back to first slide

View graphic version