Non-homogenous structures in neural networks with chaotic recursive nodes: dealing with diverse multi-assemblies architectures, connectivity and arbitrary bifurcating nodes

This paper addresses recurrent neural architectures based on bifurcating nodes that exhibit chaotic dynamics. These nodes interact through parametric coupling, self organize, and the network evolves to spatio-temporal attractors that encode stored patterns. This strategy is used to implement associative memories in which the coding of binary strings is done through period-2 cycles. The performance of such associative arrangements is measured through the average error in pattern recovery. The impact of the synaptic connections magnitude on architecture performance is analyzed, and a strategy for minimizing pattern recovery degradation when the number of stored patterns increases is developed. Experimental results show the success of such strategy. Mechanisms for allowing the studied networks to deal with asynchronous changes in input patterns, and tools for the interconnection between associative assemblies and hetero-association are developed. Finally, the coupling and coding of information in heterogeneous assemblies with diverse recursive maps are analyzed.