TRAINING FREE COUNTERPROPAGATION NETWORKS AS STATIC HETEROASSOCIATIVEMEMORIES

Counterpropagation networks (CPN) that belong to the category of self organization networks, function as statistically optimal self programm ing look up table. However, the weight adjustment criterion between th e input and the competitive layer that follows Kohonen's unsupervised learning rule, and the same between the competition and the interpolat ion layer which follows Grossberg's supervised learning rule, can be d ispensed with, without affecting the learning ability of the network. Such a network, termed as Training free counterpropagation network (TF CPN) is computationally efficient. Also, associative memories which ar e a class of neural networks, are content addressable and possess the capability to store a large set of patterns as memories. In this paper , the TFCPN scheme and its behaviour as a static hetero-associative me mory model is discussed. The TFCPN model has been successfully applied to an example in pattern classification and in the design of fink tru ss in the field of structural engineering.