Prediction Models That Learn to Avoid Missing Values
We introduce missingness-avoiding machine learning—a general framework for training models that are unlikely to rely on features that may be missing at test time.
May 6, 2025
How Should We Represent History in Interpretable Models of Clinical Policies?
Summarizing patient history through aggregation and truncation allows for interpretable and accurate modeling of policies in sequential decision-making.
Nov 2, 2024
Case-Based Off-Policy Evaluation Using Prototype Learning
We estimate the behavior policy $\mu$ for OPE using prototypes, allowing us to describe differences between $\mu$ and the target policy $\pi$ and their estimated values.
May 16, 2022