Research map / Reinforcement learning / MDPs and planning
Reinforcement Learning of Risk-Constrained Policies in Markov Decision Processes
Tomáš Brázdil, Krishnendu Chatterjee, Petr Novotný, Jiří Vahala
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 |
|---|---|---|---|
| Use RAlph instead of RAMCP | Hallway 3, Δ=0.1, sim=100 · Hallway 3 | Avg payoff: 93.18 -> 281.723 | Improved |
| Use RAlph instead of RAMCP | Hallway 3, Δ=0.1, sim=100 · Hallway 3 | Risk: 0.649 -> 0.001 | Improved |
| Use RAlph instead of RAMCP | RW 1, Δ=0.05 · RW 1 | Avg payoff: 17.78 -> 23.32 | Improved |
| Use RAlph instead of RAMCP | RW 1, Δ=0.05 · RW 1 | Risk: 0.234 -> 0.035 | Improved |
| Increase RAMCP simulations per step from 100 to 500 (ablation) | Hallway 3, Δ=0.1 · Hallway 3 | Avg payoff: 93.18 -> 161.84 | Improved |
| Reduce RAlph simulations per step from 100 to 50 (ablation) | RAlph · Hallway 3, Δ=0.1 | Avg payoff: 281.723 -> 279.32 | No clear effect |
| Increase the risk threshold from Δ=0 to Δ=0.1 (ablation) | RAlph · Hallway 3 | Avg payoff: 281.169 -> 281.723 | No clear effect |
| Use risk-constrained exploration rather than unconstrained exploration | RAlph · MDP benchmarks | chance of finding promising paths | Improved |
Closest papers by meaning
- Risk-Averse Total-Reward Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- Planning and Learning in Average Risk-aware MDPs (NeurIPS 2025 · similarity 0.89)
- Solving Robust Markov Decision Processes: Generic, Reliable, Efficient (AAAI 2025 · similarity 0.89)
- Risk-averse Total-reward MDPs with ERM and EVaR (AAAI 2025 · similarity 0.89)
- Flipping-based Policy for Chance-Constrained Markov Decision Processes (NeurIPS 2024 · similarity 0.89)
- Efficient Exploration in Average-Reward Constrained Reinforcement Learning: Achieving Near-Optimal Regret With Posterior Sampling (ICML 2024 · similarity 0.89)
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.
