Fuzzy Criteria for Feature Selection


Speaker


Abstract

Real-world optimization problems often deal with multiple different criteria, which are in most cases conflicting and/or contradictory. The use of fuzzy decision making may improve the performance of this type of systems, since it allows for an easier and suitable description of the confluence of the different criteria. For example, in the feature selection problem, one often tries to satisfy multiple criteria such as feature discriminating power, model performance or subset cardinality, which are contradictory goals. Therefore, a multi-objective formulation of the feature selection problem is more appropriate. The use of fuzzy criteria in feature selection by using a fuzzy decision making framework allows a more flexible definition of the goals, and avoids the problem of weighting different goals is classical multi-objective optimization functions. Another good example are logistic scheduling problems, which are again multi-criteria optimization problems which have many contradictory objectives and constraints, and cannot be properly described by conventional cost functions. The use of fuzzy weighted aggregation to formulate logistic problems improves the scheduling results whatever optimization methodology is used.