Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Reward-Biased Maximum Likelihood Estimation for Neural Contextual Bandits: A Distributional Learning Perspective
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 |
|---|---|---|---|
| Use NeuralRBMLE-GA instead of benchmark algorithms | Neural contextual bandits · Adult, Covertype, Magic Telescope, Mushroom, Shuttle, and MNIST | cumulative regret | Mixed |
| Use NeuralRBMLE-PC instead of benchmark algorithms | Neural contextual bandits · Adult, Covertype, Magic Telescope, Mushroom, Shuttle, and MNIST | cumulative regret | No clear effect |
| Use reward-biased estimation rather than the other benchmark methods | NeuralRBMLE and LinRBMLE · Real-world dataset simulations across random seeds | standard deviation of final regret | Helped |
| Use a categorical surrogate likelihood instead of Gaussian or Bernoulli likelihood (ablation) | NeuralRBMLE-GA · Real-world dataset experiments | empirical regret | Helped |
| Vary the reward-bias schedule to α(t) = tn, with n in {0, 0.25, 0.5, 1, 2} (ablation) | NeuralRBMLE-GA · MNIST | empirical regret | Mixed |
| Reduce hidden-layer width (ablation) | NeuralRBMLE · Real-world dataset experiments | empirical regret | No clear effect |
| Use NeuralRBMLE-GA rather than NeuralUCB or NeuralTS | Neural contextual bandits · Covertype | computation time per step (tstep): 1.117s -> 0.15s; 1.116s -> 0.15s | Helped |
Closest papers by meaning
- Exploration Through Reward Biasing: Reward-Biased Maximum Likelihood Estimation for Stochastic Multi-Armed Bandits (ICML 2020 · similarity 0.92)
- Reward-Biased Maximum Likelihood Estimation for Linear Stochastic Bandits (AAAI 2021 · similarity 0.90)
- Neural Contextual Bandits with UCB-based Exploration (ICML 2020 · similarity 0.89)
- Robustly Improving Bandit Algorithms with Confounded and Selection Biased Offline Data: A Causal Approach (AAAI 2024 · similarity 0.87)
- No-Regret Bandit Exploration based on Soft Tree Ensemble Model (NeurIPS 2024 · similarity 0.86)
- Neural Dueling Bandits: Preference-Based Optimization with Human Feedback (ICLR 2025 · 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.