Research map / Reinforcement learning / MDPs and planning
Structure Learning-Based Task Decomposition for Reinforcement Learning in Non-stationary Environments
Honguk Woo, Gwangpyo Yoo, Minjong Yoo
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 TDE+OGD instead of LILAC | RL in 2D navigation and Minitaur · 2D navigation and Minitaur | average episodic reward: 46.4% | Improved |
| Use TDE+OGD instead of LILAC | Airsim drone flying · Airsim-based drone flying | average episodic reward: 67.4% | Improved |
| Add CycleGAN-based task decomposition to the auto-encoder variant (ablation) | TDE+OGD · 2D navigation, low task uncertainty | episodic reward: 6.2% | Improved |
| Add CycleGAN-based task decomposition to the auto-encoder variant (ablation) | TDE+OGD · 2D navigation, medium and high task uncertainty | episodic reward: 14.4% (mid) and 18.6% (high) | Improved |
| Add OGD to TDE (ablation) | TDE+OGD · 2D navigation, medium and high task uncertainty | episodic reward: 8.3% (mid) and 10.1% (high) | Improved |
| Use TDE+OGD rather than TDE or TE+OGD (ablation) | RL in 2D navigation · 2D navigation, high task uncertainty | episodic reward: 14.4% (mid) and 18.6% (high); 8.3% (mid) and 10.1% (high) | Improved |
| Increase task uncertainty from low to high | TDE+OGD · 2D navigation | episodic reward: 17.3% | Hurt |
| Increase task uncertainty from low to high | LILAC · 2D navigation | episodic reward: 62.1% | Hurt |
| Increase task uncertainty from low to high | TDE+OGD · Airsim-based drone flying | episodic reward: 8.6% | Hurt |
| Increase task uncertainty from low to high | LILAC · Airsim-based drone flying | episodic reward: 12.28% | Hurt |
Closest papers by meaning
- T4NMTD: Transition-Centric Reinforcement Learning for Non-Markovian Task Decomposition (AAAI 2026 · similarity 0.88)
- Task-Agnostic Online Reinforcement Learning with an Infinite Mixture of Gaussian Processes (NeurIPS 2020 · similarity 0.88)
- Wavelet Predictive Representations for Non-Stationary Reinforcement Learning (ICLR 2026 · similarity 0.88)
- Overcoming Slow Decision Frequencies in Continuous Control: Model-Based Sequence Reinforcement Learning for Model-Free Control (ICLR 2025 · similarity 0.87)
- Prioritized Soft Q-Decomposition for Lexicographic Reinforcement Learning (ICLR 2024 · similarity 0.86)
- Periodic agent-state based Q-learning for POMDPs (NeurIPS 2024 · 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.
