Liquid state machine built of Hodgkin-Huxley neurons-pattern recognition and informational entropy

Neural networks built of Hodgkin-Huxley neurons are examined. Such structures behave like Liquid State Machines. They can effectively process geometrical patterns shown to " artificial retina " into precisely defined output. The analysis of output responses is performed in two ways: by means of Artificial Neural Network and by calculating informational entropy. 1. Introduction and problem statement The idea of simulating the behaviour of whole brain was suggested by Maass and since then it has been called Liquid State Machine (LSM) [1,2]. In general, the brain (or a fragment of it) is treated as a liquid. Neural microcircuits turn out to be very good " liquids " for computing on perturbations because of the large diversity of their elements, neurons and synapses [3], and the large variety of mechanisms and time constants characterising their interactions, involving recurrent connections on multiple spatial scales [1]. Like Turing machine, the model of LSM is based on strict mathematical framework that guarantees, under ideal conditions, universal computational power as proved in [1]. Idea of the Maass' LSM is shown in Fig. 1. The " liquid " is represented by the column consisting of some integrate and fire neurons. Randomly chosen neurons of the liquid are stimulated by input signals u's. From a formal point of view one can talk about some mapping function realised in the liquid. This function transforms the input into the " readout " layer which gives output signals y's.