BLOCK-STRUCTURED RECURRENT NEURAL NETWORKS

This paper introduces a new class of dynamic multilayer perceptrons, c alled Block Feedback Neural Networks (B(F)N). B(F)N have been develope d to provide a systematic way to build networks of high complexity, in cluding networks with coupled loops, nested loops, and so on. B(F)Ns a re specified using a block notation. Any B(F)N can be seen as a block and connected to other B(F)Ns using a fixed number of elementary conne ctions. The result of such a connection can also be considered as a bl ock, and connected to other blocks, in a recursive fashion. We develop a cost-minimizing, supervised training algorithm for this model. The algorithm is a gradient-descent type, and is tailored on the block str ucture of the model. Finally, we present some experiments of use of B( F)Ns.