EMERGENCE OF CLUSTERS IN THE HIDDEN LAYER OF A DYNAMIC RECURRENT NEURAL-NETWORK

The neural integrator of the oculomotor system is a privileged field f or artificial neural network simulation. In this paper, we were intere sted in an improvement of the biologically plausible features of the A rnold-Robinson network. This improvement was done by fixing the sign o f the connection weights in the network (in order to respect the biolo gical Dale's Law). We also introduced a notion of distance in the netw ork in the form of transmission delays between its units. These modifi cations necessitated the introduction of a general supervisor in order to train the network to act as a leaky integrator. When examining the lateral connection weights of the hidden layer, the distribution of t he weights values was found to exhibit a conspicuous structure: the hi gh-value weights were grouped in what we call clusters. Other zones ar e quite flat and characterized by low-value weights. Clusters are defi ned as particular groups of adjoining neurons which have strong and pr ivileged connections with another neighborhood of neurons. The cluster s of the trained network are reminiscent of the small clusters or patc hes that have been found experimentally in the nucleus prepositus hypo glossi, where the neural integrator is located. A study was conducted to determine the conditions of emergence of these clusters in our netw ork: they include the fixation of the weight sign, the introduction of a distance, and a convergence of the information from the hidden laye r to the motoneurons. We conclude that this spontaneous emergence of c lusters in artificial neural networks, performing a temporal integrati on, is due to computational constraints, with a restricted space of so lutions. Thus, information processing could induce the emergence of it erated patterns in biological neural networks.