FORMATION AND DYNAMICS OF MODULES IN A DUAL-TASKING MULTILAYER FEEDFORWARD NEURAL-NETWORK

We study a feed-forward neural network for two independent function ap proximation tasks. Upon training, two modules are automatically formed in the hidden layers, each handling one of the tasks predominantly. W e demonstrate that the sizes of the modules can be dynamically driven by varying the complexities of the tasks. The network serves as a simp le example of an artificial neural network with an adaptable modular s tructure. This Study was motivated by related dynamical nature of modu les in animal brains.