Towards federated feature selection: Logarithmic division for resource-conscious methods
Use este enlace para citar
http://hdl.handle.net/2183/38025
Excepto si se señala otra cosa, la licencia del ítem se describe como Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC-BY-NC-ND 4.0)
Colecciones
- GI-LIDIA - Artigos [65]
Metadatos
Mostrar el registro completo del ítemTítulo
Towards federated feature selection: Logarithmic division for resource-conscious methodsFecha
2024Cita bibliográfica
Suárez-Marcote, S., Morán-Fernández, L., & Bolón-Canedo, V. (2024). Towards federated feature selection: Logarithmic division for resource-conscious methods. Neurocomputing, 128099. https://doi.org/10.1016/j.neucom.2024.128099
Resumen
[Abstract]: Feature selection is a popular preprocessing step to reduce the dimensionality of the data while preserving the important information. In this paper, we propose an efficient and green feature selection method based on information theory, with the novelty of using the logarithmic division and resorting to fixed-point precision. Moreover, we extend these advancements by adapting the Mutual Information calculation for federated scenarios. The results of experiments conducted on several datasets indicate the potential of our proposal, as it does not incur significant information loss compared to the double-precision method, both in the features selected and in the subsequent classification step. Our method has shown significant potential when applied in federated scenarios, where experimentation demonstrated a lossless feature selection process and maintains classification results compared with centralised versions. These findings open up possibilities towards a new family of green feature selection methods, which would help to minimise energy consumption, lower carbon emissions and increase adaptability to Internet of Things environments.
Palabras clave
Feature selection
Mutual information
Low precision
Internet of things
Federated learning
Logarithmic division
Mutual information
Low precision
Internet of things
Federated learning
Logarithmic division
Versión del editor
Derechos
Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC-BY-NC-ND 4.0)
ISSN
0925-2312 (print)
1872-8286 (electronic)
1872-8286 (electronic)