Learning Pareto-Efficient Decisions with Confidence

Sofia Ek, Dave Zachariah, Peter Stoica
Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, PMLR 151:9969-9981, 2022.

Abstract

The paper considers the problem of multi-objective decision support when outcomes are uncertain. We extend the concept of Pareto-efficient decisions to take into account the uncertainty of decision outcomes across varying contexts. This enables quantifying trade-offs between decisions in terms of tail outcomes that are relevant in safety-critical applications. We propose a method for learning efficient decisions with statistical confidence, building on results from the conformal prediction literature. The method adapts to weak or nonexistent context covariate overlap and its statistical guarantees are evaluated using both synthetic and real data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v151-ek22a, title = { Learning Pareto-Efficient Decisions with Confidence }, author = {Ek, Sofia and Zachariah, Dave and Stoica, Peter}, booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics}, pages = {9969--9981}, 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/ek22a/ek22a.pdf}, url = {https://proceedings.mlr.press/v151/ek22a.html}, abstract = { The paper considers the problem of multi-objective decision support when outcomes are uncertain. We extend the concept of Pareto-efficient decisions to take into account the uncertainty of decision outcomes across varying contexts. This enables quantifying trade-offs between decisions in terms of tail outcomes that are relevant in safety-critical applications. We propose a method for learning efficient decisions with statistical confidence, building on results from the conformal prediction literature. The method adapts to weak or nonexistent context covariate overlap and its statistical guarantees are evaluated using both synthetic and real data. } }
Endnote
%0 Conference Paper %T Learning Pareto-Efficient Decisions with Confidence %A Sofia Ek %A Dave Zachariah %A Peter Stoica %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-ek22a %I PMLR %P 9969--9981 %U https://proceedings.mlr.press/v151/ek22a.html %V 151 %X The paper considers the problem of multi-objective decision support when outcomes are uncertain. We extend the concept of Pareto-efficient decisions to take into account the uncertainty of decision outcomes across varying contexts. This enables quantifying trade-offs between decisions in terms of tail outcomes that are relevant in safety-critical applications. We propose a method for learning efficient decisions with statistical confidence, building on results from the conformal prediction literature. The method adapts to weak or nonexistent context covariate overlap and its statistical guarantees are evaluated using both synthetic and real data.
APA
Ek, S., Zachariah, D. & Stoica, P.. (2022). Learning Pareto-Efficient Decisions with Confidence . Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 151:9969-9981 Available from https://proceedings.mlr.press/v151/ek22a.html.

Related Material