Feature Selection Scheme Based on Zero-Sum Two-Player Game

Abstract

We propose a new filter methodology for feature selection using the concept of game theory whereby features are assimilated to players. In this game theoretical context, a strategy corresponds to a particular affinity between a group of features forming a cluster, and the payoff function is computed based on the weighted distance between a feature and a cluster. A zerosum two-player game problem is solved through a global combination of pairwise features. Finally, each feature is represented by the value of the objective function, at the optimal solution, which indicates the contribution of each feature. The importance of features is then evaluated by their optimal values. To validate the effectiveness of the proposed methodology, we have conducted a classification task utilizing SVM on various UCI and statlog datasets. The experimental results show that the proposed scheme leads to improvement in classification performance, when compared to mRMR and Fisher score algorithms.

Determination of an optimal feature selection method based on maximum Shapley value

“…This paper focuses on the first issue, namely selecting input variables in an attempt to maximize the perfonnance of a classifier on previously unseen data. It is worth noting that a large number of algorithms have been proposed in the literature for feature subset selection [3][4][5][6][7][8]. However, most of the proposed methodologies exhibit a common drawback: The selected features depend strongly on the application at hand, and there is no general feature selection scheme that provides an optimal classification with different data (e.g.…”