CUDA acceleration of MI-based feature selection methods

• Published in: Journal of parallel and distributed computing Vol. 190; p. 104901
• Main Authors: Beceiro, Bieito, González-Domínguez, Jorge, Morán-Fernández, Laura, Bolón-Canedo, Verónica, Touriño, Juan
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01-08-2024
• Subjects: CUDA; Feature selection; Fixed point; Low precision; Mutual information; CUDA; Low precision; Feature selection; Mutual information; Fixed point
• ISSN: 0743-7315; 1096-0848
• DOI: 10.1016/j.jpdc.2024.104901

Feature selection algorithms are necessary nowadays for machine learning as they are capable of removing irrelevant and redundant information to reduce the dimensionality of the data and improve the quality of subsequent analyses. The problem with current feature selection approaches is that they are computationally expensive when processing large datasets. This work presents parallel implementations for Nvidia GPUs of three highly-used feature selection methods based on the Mutual Information (MI) metric: mRMR, JMI and DISR. Publicly available code includes not only CUDA implementations of the general methods, but also an adaptation of them to work with low-precision fixed point in order to further increase their performance on GPUs. The experimental evaluation was carried out on two modern Nvidia GPUs (Turing T4 and Ampere A100) with highly satisfactory results, achieving speedups of up to 283x when compared to state-of-the-art C implementations.
•MI-based feature selection is expensive and unfeasible for huge datasets.•A parallel CUDA version for JMI, DISR and mRMR is proposed.•Several optimizations are included to improve memory accesses.•The CUDA implementations efficiently exploit the hardware of modern Nvidia GPUs.