Feature selection in machine learning: Rényi min-entropy vs Shannon entropy
This work addresses feature selection for machine learning practitioners, offering an incremental improvement by exploring an alternative entropy measure.
The paper tackles feature selection in machine learning by comparing Rényi min-entropy with Shannon entropy, showing that the Rényi-based algorithm outperforms the Shannon-based one on real datasets like BASEHOCK, SEMEION, and GISETTE.
Feature selection, in the context of machine learning, is the process of separating the highly predictive feature from those that might be irrelevant or redundant. Information theory has been recognized as a useful concept for this task, as the prediction power stems from the correlation, i.e., the mutual information, between features and labels. Many algorithms for feature selection in the literature have adopted the Shannon-entropy-based mutual information. In this paper, we explore the possibility of using Rényi min-entropy instead. In particular, we propose an algorithm based on a notion of conditional Rényi min-entropy that has been recently adopted in the field of security and privacy, and which is strictly related to the Bayes error. We prove that in general the two approaches are incomparable, in the sense that we show that we can construct datasets on which the Rényi-based algorithm performs better than the corresponding Shannon-based one, and datasets on which the situation is reversed. In practice, however, when considering datasets of real data, it seems that the Rényi-based algorithm tends to outperform the other one. We have effectuate several experiments on the BASEHOCK, SEMEION, and GISETTE datasets, and in all of them we have indeed observed that the Rényi-based algorithm gives better results.