Feature Collapsing for Gaussian Process Variable Ranking

Isaac Sebenius, Topi Paananen, Aki Vehtari
Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, PMLR 151:11341-11355, 2022.

Abstract

At present, there is no consensus on the most effective way to establish feature relevance for Gaussian process models. The most common heuristic, Automatic Relevance Determination, has several downsides; many alternate methods incur unacceptable computational costs. Existing methods based on sensitivity analysis of the posterior predictive distribution are promising, but are heavily biased and show room for improvement. This paper proposes Feature Collapsing as a novel method for performing GP feature relevance determination in an effective, unbiased, and computationally-inexpensive manner compared to existing algorithms.

Cite this Paper

BibTeX
@InProceedings{pmlr-v151-sebenius22a, title = { Feature Collapsing for Gaussian Process Variable Ranking }, author = {Sebenius, Isaac and Paananen, Topi and Vehtari, Aki}, booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics}, pages = {11341--11355}, year = {2022}, editor = {Camps-Valls, Gustau and Ruiz, Francisco J. R. and Valera, Isabel}, volume = {151}, series = {Proceedings of Machine Learning Research}, month = {28--30 Mar}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v151/sebenius22a/sebenius22a.pdf}, url = {https://proceedings.mlr.press/v151/sebenius22a.html}, abstract = { At present, there is no consensus on the most effective way to establish feature relevance for Gaussian process models. The most common heuristic, Automatic Relevance Determination, has several downsides; many alternate methods incur unacceptable computational costs. Existing methods based on sensitivity analysis of the posterior predictive distribution are promising, but are heavily biased and show room for improvement. This paper proposes Feature Collapsing as a novel method for performing GP feature relevance determination in an effective, unbiased, and computationally-inexpensive manner compared to existing algorithms. } }
Endnote
%0 Conference Paper %T Feature Collapsing for Gaussian Process Variable Ranking %A Isaac Sebenius %A Topi Paananen %A Aki Vehtari %B Proceedings of The 25th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2022 %E Gustau Camps-Valls %E Francisco J. R. Ruiz %E Isabel Valera %F pmlr-v151-sebenius22a %I PMLR %P 11341--11355 %U https://proceedings.mlr.press/v151/sebenius22a.html %V 151 %X At present, there is no consensus on the most effective way to establish feature relevance for Gaussian process models. The most common heuristic, Automatic Relevance Determination, has several downsides; many alternate methods incur unacceptable computational costs. Existing methods based on sensitivity analysis of the posterior predictive distribution are promising, but are heavily biased and show room for improvement. This paper proposes Feature Collapsing as a novel method for performing GP feature relevance determination in an effective, unbiased, and computationally-inexpensive manner compared to existing algorithms.
APA
Sebenius, I., Paananen, T. & Vehtari, A.. (2022). Feature Collapsing for Gaussian Process Variable Ranking . Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 151:11341-11355 Available from https://proceedings.mlr.press/v151/sebenius22a.html.

Related Material