Research map / Reinforcement learning / MDPs and planning
Noise as a Natural Regularizer in Markov Decision Processes: Connecting Environmental Stochasticity and Policy Simplicity
Harry Chen, Yiyang Sun, Michal Moshkovitz, Zachery Boner, Lesia Semenova, Cynthia Rudin, Ronald Parr
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add reset noise to the hallway MDP versus lowering the discount factor | optimal policy · Hallway grid MDP, reset noise alpha=0.95 | policy distance between clean and noisy optimal policies | Improved |
| add uniform action noise (epsilon 0.3) versus lowering discount | optimal policy · Hallway grid MDP | policy distance | Improved |
| add sticky noise | optimal policy · Hallway grid MDP | policy equivalence | Improved |
| add catastrophic-risk transitions (Tightrope MDP) so noise cannot be mimicked by discounting | optimal policy · Tightrope MDP | optimal action choice | Hurt |
| increase random action noise probability with gamma fixed at 0.99 | PPO policy · 2-Link Reacher | trajectory stability | Hurt |
| lower the discount factor | PPO policy · 2-Link Reacher | reaching precision | Hurt |
| compare injected noise to reduced discount via reward-distribution JSD | PPO policy · 2-Link Reacher | JSD of reward distributions | Improved |
Closest papers by meaning
- Horizon-free Learning for Markov Decision Processes and Games: Stochastically Bounded Rewards and Improved Bounds (ICML 2023 · similarity 0.90)
- Beyond Average Return in Markov Decision Processes (NeurIPS 2023 · similarity 0.89)
- Non-Stationary Markov Decision Processes, a Worst-Case Approach using Model-Based Reinforcement Learning (NeurIPS 2019 · similarity 0.89)
- Reinforcement Learning with Random Time Horizons (ICML 2025 · similarity 0.88)
- On Shallow Planning Under Partial Observability (AAAI 2025 · similarity 0.88)
- On the Convergence of Success Conditioning for Policy Optimization (NeurIPS 2026 · similarity 0.87)
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.
