Multi-Layer Perceptrons

• Proposed mid-1960s, but learning algorithms not available until mid-1980s

• Continuous inputs

• Continuous outputs

• Continuous activation functions used

• e.g. sigmoid, tanh

• Feed forward networks