Periodic Signal Learning and Recognition in Coupled Oscillators

Abstract

Responses to periodic signals in the oscillator network are studied numerically. The network used here is globally coupled oscillators with distributed natural frequencies of the oscillatory elements. The ability of the temporal-pattern recognition in the system is studied using the learning rule which changes the distribution of the natural frequencies. It is found that the coherency representing the order of the system exhibits the periodic and complex behaviors. The periodic response of the coherency is observed in the case where the input signal is similar to the learned signal. The behavior of the coherency becomes complex; i.e., the number of the degrees of freedom of the behavior increases with the decrease in the similarity of the input signal to the learned signal. Moreover, the long transient process is observed after turning off the input.