Proceedings chapter
Open access

Structured nonlinear variable selection

Published inConference on Uncertainty in Artificial Intelligence, UAI2018, Editors Globerson, Amir & Silva, Ricardo
Presented at Monterey, CA (USA), 6-10 August 2018
PublisherRed Hook, N.Y. : Curran Associates
Publication date2018

We investigate structured sparsity methods for variable selection in regression problems where the target depends nonlinearly on the inputs. We focus on general nonlinear functions not limiting a priori the function space to additive models. We propose two new regularizers based on partial derivatives as nonlinear equivalents of group lasso and elastic net. We formulate the problem within the framework of learning in reproducing kernel Hilbert spaces and show how the variational problem can be reformulated into a more practical finite dimensional equivalent. We develop a new algorithm derived from the ADMM principles that relies solely on closed forms of the proximal operators. We explore the empirical properties of our new algorithm for Nonlinear Variable Selection based on Derivatives (NVSD) on a set of experiments and confirm favourable properties of our structured-sparsity models and the algorithm in terms of both prediction and variable selection accuracy.

  • arxiv : stat.ML
  • European Commission - Road-, Air- and Water-based Future Internet Experimentation [645220]
Citation (ISO format)
GREGOROVA, Magda, KALOUSIS, Alexandros, MARCHAND-MAILLET, Stéphane. Structured nonlinear variable selection. In: Conference on Uncertainty in Artificial Intelligence, UAI2018. Monterey, CA (USA). Red Hook, N.Y. : Curran Associates, 2018.
Main files (1)
Proceedings chapter (Accepted version)

Technical informations

Creation01/15/2020 4:17:00 PM
First validation01/15/2020 4:17:00 PM
Update time03/15/2023 6:46:40 PM
Status update03/15/2023 6:46:40 PM
Last indexation01/17/2024 8:48:29 AM
All rights reserved by Archive ouverte UNIGE and the University of GenevaunigeBlack