Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions
By Omer Gottesman, Joseph D. Futoma, Yao Liu, S. Parbhoo, Leo Anthony Celi, Emma Brunskill, Finale Doshi-Velez. 2020
69 citations recorded by the source corpus.
Off-policy evaluation in reinforcement learning offers the chance of using observational data to improve future outcomes in domains such as healthcare and education, but safe deployment in high stakes settings requires ways of assessing its validity. Traditional measures such as confidence intervals may be insufficient due to noise, limited data and confounding. In this paper we develop a method that could serve as a hybrid human-AI system, to enable human experts to analyze the validity of policy evaluation estimates. This is accomplished by highlighting observations in the data whose removal will have a large effect on the OPE estimate, and formulating a set of rules for choosing which ones to present to domain experts for validation. We develop methods to compute exactly the influence functions for fitted Q-evaluation with two different function classes: kernel-based and linear least squares, as well as importance sampling methods. Experiments on medical simulations and real-world intensive care unit data demonstrate that our method can be used to identify limitations in the evaluation process and make evaluation more robust.
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Pratik Chaudhari
- Rasool Fakoor
- Jonas W. Mueller
- Alex Smola
- Finale Doshi-Velez
- Joseph D. Futoma
- Jianzhun Du
- Yoshua Bengio
Citations in the corpus
- Continuous Doubly Constrained Batch Reinforcement Learning (2021, background): “, sensitive healthcare applications [19].”
- Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs (2020, background): “, in ICU [Gottesman et al., 2020]) might benefit from this in particular, because practitioners usually assume the presence of a large batch of already-collected data, which consists of patient measurements with irregular observations and actions.”