Preventing failures due to dataset shift: Learning predictive models that transport

Adarsh Subbaswamy, Peter Schulam, Suchi Saria

Research output: Contribution to conferencePaper

Abstract

Classical supervised learning produces unreliable models when training and target distributions differ, with most existing solutions requiring samples from the target domain. We propose a proactive approach which learns a relationship in the training domain that will generalize to the target domain by incorporating prior knowledge of aspects of the data generating process that are expected to differ as expressed in a causal selection diagram. Specifically, we remove variables generated by unstable mechanisms from the joint factorization to yield the Surgery Estimator-an interventional distribution that is invariant to the differences across environments. We prove that the surgery estimator finds stable relationships in strictly more scenarios than previous approaches which only consider conditional relationships, and demonstrate this in simulated experiments. We also evaluate on real world data for which the true causal diagram is unknown, performing competitively against entirely data-driven approaches.

Original languageEnglish (US)
StatePublished - 2020
Event22nd International Conference on Artificial Intelligence and Statistics, AISTATS 2019 - Naha, Japan
Duration: Apr 16 2019Apr 18 2019

Conference

Conference22nd International Conference on Artificial Intelligence and Statistics, AISTATS 2019
CountryJapan
CityNaha
Period4/16/194/18/19

ASJC Scopus subject areas

  • Artificial Intelligence
  • Statistics and Probability

Fingerprint Dive into the research topics of 'Preventing failures due to dataset shift: Learning predictive models that transport'. Together they form a unique fingerprint.

  • Cite this

    Subbaswamy, A., Schulam, P., & Saria, S. (2020). Preventing failures due to dataset shift: Learning predictive models that transport. Paper presented at 22nd International Conference on Artificial Intelligence and Statistics, AISTATS 2019, Naha, Japan.