Analyzing and inferring distance metrics on the particle competition and cooperation algorithm

Abstract

Machine Learning is an increasing area over the last few years and it is one of the highlights in Artificial Intelligence area. Nowadays, one of the most studied areas is Semi-supervised learning, mainly due to its characteristic of lower cost in labeling sample data. The most active category in this subarea is that of graph-based models. The Particle Competition and Cooperation in Networks algorithm is one of the techniques in this field, which has always used the Euclidean distance to measure the similarity between data and to build the graph. This project aims to implement the algorithm and apply other distance metrics in it, over different datasets. Thus, the results on these metrics are compared to analyze if there is such a metric that produces better results, or if different datasets require a different metric in order to obtain a better correct classification rate. We also expand this gained knowledge, proposing how to identify the best metric for the algorithm based on its initial graph structure, with no need to run the algorithm for each metric we want to evaluate.