Abstract—Constructive Neural Network learning algorithms provide incremental ways to determine the near-minimal architecture of a multi layer perceptron network along with learning algorithms for determining appropriate weights for pattern classification problems. An improved version of adaptive learning algorithm in a structured multilayer networks is proposed in this research work. A proper weight setting for the constructive architecture to solve pattern classification problems is analyzed and tabulated.

Index Terms—Adaptive Resonance Theory, Constructive Neural Network, Pattern Classification.

Cite: S.S.Sridhar and M.Ponnavaikko, "Improved Adaptive Learning Algorithm for Constructive Neural Networks," International Journal of Computer and Electrical Engineering vol. 3, no. 1, pp. 30-36, 2011.