Research map / Reinforcement learning / Visual and representation RL
T4NMTD: Transition-Centric Reinforcement Learning for Non-Markovian Task Decomposition
Ruixuan Miao, Xu Lu, Cong Tian, Bin Yu, Zhenhua Duan
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 T4NMTD instead of baselines (HRM, QRM, LSTS, HDQN, MOD, DIRL) on sample efficiency | T4NMTD · Tasks 1-8 | interactions / success rate: Task7: 4.4 interactions (x1e6) at 100% vs HRM 19.1 at 10% | Improved |
| T4NMTD vs MOD on Task1 interactions | T4NMTD · Task1 | interactions (x1e6): 0.8±0.1 vs 0.8±0.1 | No clear effect |
| replace proposition-based decomposition with random partition of DFA transitions (ablation) | T4NMTD · Waterworld/Halfcheetah/Racecar/Frozenlake tasks | convergence rate | Improved |
| add dynamic policy intervention (value propagation across sub-learners) (ablation) | T4NMTD · all NMTs | reward / convergence | Improved |
| reuse trained low-level sub-learners when transferring to a new NMT | T4NMTD · four transfer scenarios | convergence rate: over 2.67x faster | Improved |
Closest papers by meaning
- Prioritized Soft Q-Decomposition for Lexicographic Reinforcement Learning (ICLR 2024 · similarity 0.88)
- Structure Learning-Based Task Decomposition for Reinforcement Learning in Non-stationary Environments (AAAI 2022 · similarity 0.88)
- Unifying Task Specification in Reinforcement Learning (ICML 2017 · similarity 0.88)
- TTRL: Test-Time Reinforcement Learning (NeurIPS 2025 · similarity 0.87)
- Temporal Distance-aware Transition Augmentation for Offline Model-based Reinforcement Learning (ICML 2025 · similarity 0.86)
- Unveiling Markov heads in Pretrained Language Models for Offline Reinforcement Learning (ICML 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.
