Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Counterfactual Risk Minimization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace standard L2 regularization with logging-policy regularization in IPS | IPS-L2 · Fashion-MNIST test | expected reward (stoch.): 0.7778 -> 0.7955 | Helped |
| Replace standard L2 regularization with logging-policy regularization in IPS | IPS-L2 · CIFAR-100 test | expected reward (stoch.): 0.3475 -> 0.5553 | Helped |
| Use weighted negative log-likelihood with logging-policy regularization instead of variance regularization | POEM · Fashion-MNIST test | expected reward (stoch.): 0.8060 -> 0.7978 | No clear effect |
| Use weighted negative log-likelihood with logging-policy regularization instead of variance regularization | POEM · CIFAR-100 test | expected reward (stoch.): 0.3338 -> 0.6143 | Helped |
| Estimate the logging policy from logged actions and use it for LPR (ablation) | IPS-LPR · Fashion-MNIST test | expected reward (stoch.): 0.7955 -> 0.7950 | No clear effect |
| Estimate the logging policy from logged actions and use it for LPR (ablation) | WNLL-LPR · CIFAR-100 test | expected reward (stoch.): 0.6143 -> 0.6143 | No clear effect |
| Make the logging policy more uniform (ablation) | IPS-LPR relative to IPS-L2 · Fashion-MNIST simulated logging policies | expected test reward | Mixed |
| Use logging-policy regularization when the logging policy is worse than uniform (ablation) | IPS-LPR relative to IPS-L2 · simulated logging policy worse than uniform | expected test reward | Hurt |
| Use variance regularization instead of LPR | IPS-LPR · Fashion-MNIST training | training time: 25% | Hurt |
Closest papers by meaning
- Sequential Counterfactual Risk Minimization (ICML 2023 · similarity 0.92)
- Efficient Counterfactual Learning from Bandit Feedback (AAAI 2019 · similarity 0.91)
- Bayesian Regret Minimization in Offline Bandits (ICML 2024 · similarity 0.89)
- Logarithmic Smoothing for Pessimistic Off-Policy Evaluation, Selection and Learning (NeurIPS 2024 · similarity 0.87)
- Log-Sum-Exponential Estimator for Off-Policy Evaluation and Learning (ICML 2025 · similarity 0.86)
- Regret minimization in Linear Bandits with offline data via extended D-optimal exploration. (NeurIPS 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.