Design of a multi-agent classification system

In the era of information, emergent paradigms such as the Internet of Things with billions of devices constantly connected to the Internet exchanging heterogeneous data, demand new computing approaches for intelligent big data processing. Given this technological scenario, a suitable approach leverages on many computational entities, each of which performs tasks with low computational burden, conceived to be executed on multi-core and many-core architectures. To this aim, in this thesis, we propose Evolutive Agent Based Clustering (E-ABC) as promising framing reference. E-ABC is conceived to orchestrate a swarm of intelligent agents acting as individuals of an evolving population, each performing a random walk on a different subset of patterns. Each agent is in charge of discovering well-formed (compact and populated) clusters and, at the same time, a suitable subset of features corresponding to the subspace where such clusters lie, following a local metric learning approach, where each cluster is characterized by its own subset of relevant features. E-ABC is able to process data belonging to structured and possibly non-metric spaces, relying on custom parametric dissimilarity measures. Specifically, two variants are investigated. A first variant, namely E-ABC, aims at solving unsupervised problems, where agents’ task is to find well-formed clusters lying in suitable subspaces. A second variant, E-ABC^2, aims at solving classification problems by synthesizing a classification system on the top of the clusters discovered by the swarm. In particular, as a practical and real-world application, this novel classification system has been employed for recognizing and predicting localized faults on the electric distribution network of Rome, managed by the Italian utility company ACEA. Tests results show that E-ABC is able to synthesize classification models characterized by a remarkable generalization capability, with adequate performances to be employed in Smart Grids condition-based management systems. Moreover, the feature subsets where most of the meaningful clusters have been discovered can be used to better understand sub-classes of failures, each identified by a set of related causes.