Neural Network Architectures and Learning Schemes

Abstract

Existing neural network architectures can be divided into three basic categories: Feed forward, Feed-back, and Self-organizing neural networks. The most widely used neural architectures that can be classified into these three categories are shown in Figure 2.1. Although each of these categories is based on a different philosophy and obeys different principles, the characterization of a system by the general term “neural network” usually implies an ability to learn. Learning is the process by which a neural system acquires ability to carry out certain tasks by adjusting its internal parameters according to some learning scheme. Depending on the particular neural architecture considered, learning can be either supervised or unsupervised.