An evolutionary agents based system for data mining and local metric learning

Discovering regularities in Big Data is nowadays a crucial task in many different applications, from bioinformatics to cybersecurity. To this aim, a promising approach consists in performing data clustering with Local Metric Learning, i.e. trying to discover well-formed (compact and populated) clusters and, at the same time, a suitable subset of features corresponding to the subspace where each cluster lies. In this paper, we propose a novel evolutionary agent based clustering algorithm, where agents act as individuals of an evolving population, each one performing a simple cluster analysis task on a different subset of patterns randomly drawn from the entire dataset. A customized genetic algorithm orchestrates the evolution of a population of such agents. This approach is able to reveal clusters that are not evident when using the whole set of features, and can deal with large datasets, as each agent will process a small subset of patterns. First results obtained considering a synthetic dataset are encouraging, suggesting future developments.