Emergent Mechanism of Swarms Based on Machine Learning Approach

—Acquisition of Collective Motions Based on Group Escaping Strategy from a Predator—

Abstract

Why do animals form swarms? In general, research into such swarms is often conducted by directly examining their movements of a swarm and presenting them as a mathematical model (in terms of differential equations) by comparing them with the movements of live animals. Although there several top-down approaches for studying animal swarms, few studies have investigated the underlying reason of emergence of swarms. In this study, we attempt to investigate this via an artificial approach based on the machine learning method. A single predator and multiple escapees are used to investigate whether or not the escapees form swarms for prolonged survival, and if so, how their swarming behavior can be modeled.