Two-stage feature selection using ranking self-adaptive differential evolution algorithm for recognition of acceleration activity

Abstract

Widespread research on activity recognition is becoming an imperative topic for improving the quality ofhuman health. The fast development of sensing technology has become a fundamental platform for researchers toimplement a system that could fulfill human needs. Due to privacy interests and low cost, wearable sensing technology isused in numerous physical activity monitoring and recognition systems. While these systems have proved to be successful,it is crucial to pay attention to the less relevant features to be classified. In such circumstances, it might happen that somefeatures are less meaningful for describing the activity. Less complex and easy to understand, feature ranking is gaininga lot of attention in most feature dimension problems such as in bioinformatics and hyperspectral images. However, theimprovement of ranking features in activity recognition has not yet been achieved. On the other hand, an evolutionaryalgorithm has proven its effectiveness in searching the best feature subsets. An exhaustive searching process of findingan optimal parameter value is another challenge. Consequently, this paper proposes a ranking self-adaptive differentialevolution (rsaDE) feature selection algorithm. The proposed algorithm is capable of selecting the optimal feature subsetswhile improving the recognition of acceleration activity using a minimum number of features. The experiments employedreal-world physical acceleration data sets: WISDM and PAMAP2. As a result, rsaDE performed better than the currentmethods in terms of model performance and its efficiency in the context of random forest ensemble classifiers.