Constructions and Dynamics of Neural Networks
In this chapter, we present constructions of neural networks and their dynamics in processing, learning, and self-organization. Biological neural networks, in particular, most of interface networks, evolved in such a specific manner that each network fits its particular purpose such as seeing, listening, or speaking. Artificial neural networks also possess various constructions dependent on purposes. Therefore, in this chapter, we investigate the constructions and dynamics in individual networks according to purposes. However, there exits a common dynamics in their microscopic mechanisms of learning and self-organization, namely, the Hebbian rule in the broad sense of the word. First, we consider the Hebbian rule. Then we go on to various constructions and dynamics in networks. You do not need any detailed background in advance since we begin with conventional (real-valued) neural networks and, afterward, we extend them into complex-valued neural networks (CVNNs).
KeywordsNeural Network Hide Layer Input Signal Independent Component Analysis Associative Memory
Unable to display preview. Download preview PDF.