Distributionally Robust Policy Evaluation and Learning in Offline Contextual Bandits

Authors: Nian Si, Fan Zhang, Zhengyuan Zhou, Jose Blanchet

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Additionally, we provide extensive simulations to demonstrate the robustness of our policy.
Researcher Affiliation Collaboration 1Department of Management Science & Engineering, Stanford University 2IBM research and Stern School of Business, New York University.
Pseudocode Yes Algorithm 1 Distributionally Robust Policy Evaluation
Open Source Code No The paper does not provide any specific links or explicit statements about the release of source code.
Open Datasets No The paper describes a simulation environment for data generation but does not provide access information (link, DOI, citation) to a publicly available or open dataset. For example: 'The feature vectors Xi R10 are independently and uniformly drawn from [0, 1]10.'
Dataset Splits Yes We first test the convergence of different estimators for δ = 0.2 and three different sizes of dataset: n = 103, 104, 105.
Hardware Specification No No explicit hardware specifications (e.g., GPU/CPU models, memory details) were mentioned in the paper.
Software Dependencies No No specific software dependencies with version numbers were mentioned in the paper.
Experiment Setup Yes We fix δ = 0.1 and the size of training set is n = 3000, and the policy class is depth-3 trees.