Chaotic Time Series Prediction for Rock-Paper-Scissors using Adaptive Social Behaviour Optimization (ASBO)

Time series prediction involves analyzing a set of data from past and current occurrences in order to predict the future set of data. In dynamic systems, chaotic behaviour is intrinsically observable and the resulting chaotic time series have nonlinear characteristics. Nevertheless, such data can be optimized to make sense out of the chaos. Multiple algorithms exist to this end, which have various applications. In this paper, this phenomenon is illustrated by making use of the well-known game rock-paper-scissors(R-P-S) played between two agents, one real and one adaptive. It is possible to identify and predict patterns in the choices made by the real agent during the course of play by analyzing the sequence of chaotic data using the Adaptive Social Behaviour Optimization (ASBO) algorithm. This optimization method makes use of a self-adaptive mutation strategy which takes into consideration dynamic factors such as leadership, confidence and competition, which are all functions of time.