Improving on Perceptron Training
Steven E. Hampson · Birkhäuser Boston eBooks · 1990
When positive and negative input patterns are linearly separable, use of the perceptron training procedure guarantees convergence on the correct output. Theoretically this is sufficient, but in practice convergence may be quite slow. In this chapter, the basic time complexity of perceptron training is considered, and three techniques are developed which can significantly improve performance. These are variable associability, adaptive origin placement, and short-term weight modification. Each addresses a different limiting aspect of perceptron training.