Learning and Generalization in Random Automata Networks

Abstract. It has been shown [7,16] that feedforward Boolean networks can learn to perform specific simple tasks and generalize well if only a subset of the learning examples is provided for learning. Here, we extend this body of work and show experimentally that random Boolean net-works (RBNs), where both the interconnections and the Boolean transfer functions are chosen at random initially, can be evolved by using a state-topology evolution to solve simple tasks. We measure the learning and generalization performance, investigate the influence of the average node connectivity K, the system size N, and introduce a new measure that allows to better describe the network’s learning and generalization be-havior. Our results show that networks with higher average connectivity K (supercritical) achieve higher memorization and partial generalization. However, near critical connectivity, the networks show a higher perfect generalization on the even-odd task.