Using Machine Learning Techniques for Multi-agent Systems Testing

This paper addresses the problem of effective multi-agent systems (MAS) testing using machine learning techniques. It reviews various machine learning methods that can be used for the generation of tests in MASs, such as reinforcement learning, evolutionary algorithms, generative models and others. It presents a comprehensive review that analyze the strengths, weaknesses and the adequacy of each method to the different characteristics of MASs and test objectives. It also identifies certain open challenges and future orientations for continuing research in this area. This document aims to provide researchers and practitioners with a clear overview and advice on the selection of the ML approach most suited to their specific test needs.