Feature extraction and selection using statistical dependence criteria

TOMASSI, DIEGO; MARX, NICOLÁS; BEAUSEROY, PIERRE

Buenos Aires

Simposio; Argentine Symposium on Artificial Intelligence; 2016

SADIO

Dimensionality reduction using feature extraction and selection approaches is a common stage of many regression and classification tasks. In recent years there have been significant efforts to reduce the dimension of the feature space without lossing information that is relevant for prediction. This objective can be cast into a conditional independence condition between the response or class labels and the transformed features. Building on this, in this work we use measures of statistical dependence to estimate a lower-dimensional linear subspace of the features that retains the sufficient information. Unlike likelihood-based and many momentbased methods, the proposed approach is semi-parametric and does not require model assumptions on the data. A regularized version to achieve simultaneous variable selection is presented too. Experiments with simulated data show that the performance of the proposed method comparesfavorably to well-known linear dimension reduction techniques.