Rough-set Based Genetic Algorithms for Weakly Supervised Feature Selection

In this article, we study the problem of feature selection under weak supervision, focusing in particular on the fuzzy labels setting, where the weak supervision is provided in terms of possibility distributions over candidate labels. While traditional Rough Set-based approaches have been applied for tackling this problem, they have high computational complexity and only provide local search heuristic methods. In order to address these issues, we propose a global optimization algorithm, based on genetic algorithms and Rough Set theory, for feature selection under fuzzy labels. Based on a set of experiments, we illustrate the effectiveness of the proposed approach in comparison to state-of-the-art methods.