SS11: Soft Computing Techniques for Machine Learning

This special session is aimed at discussing recent and novel fuzzy methods to deal with the current challenges on machine learning. This research field is very active due to the large number of real-world problems that can be faced using techniques of this field. The canonical problems of this area of research are classification, regression and clustering. However, in recent years there are a great number of hot topics like the problem of imbalanced data, low quality and/or noisy instances, semi-supervised learning or multi-label and multi-instance problems among others. When tackling the previously mentioned problems, soft computing techniques are widely applied. Specifically, fuzzy systems are a common tool as they provide an interpretable model understandable by human beings whilst the results obtained are accurate, since fuzzy logic has an inherent ability to cope with the great uncertainty present in these new challenging problems. Evolutionary computation is a robust technique for optimization, learning and adaptation tasks. They can adjust the model parameters for each specific problem for the sake of enhancing their performance. The synergy between these two techniques leads to a better capability for the design and optimization of fuzzy models. Moreover, Big Data also offers new possibilities for fuzzy methods, where new challenges appear with respect to their scalability when dealing with enormous amounts of data. The special session is composed of seven contributions dealing with different topics of the machine learning field.