Research map / Reinforcement learning / Visual and representation RL
Reinforcement Learning in POMDPs With Memoryless Options and Option-Observation Initiation Sets
Denis Steckelmacher, Diederik Roijers, Anna Harutyunyan, Peter Vrancx, Hélène Plisnier, Ann Nowé
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 |
|---|---|---|---|
| Condition option initiation on the previously executed option | Options in Object Gathering · Object Gathering | cumulative reward per episode | Improved |
| Use LSTM memory over options instead of OOIs | Object Gathering agent · Object Gathering | episodes needed to learn a good policy | Hurt |
| Condition option initiation on the previously executed option | Options in Modified DuplicatedInput · Modified DuplicatedInput | cumulative reward per episode | Improved |
| Use standard initiation sets instead of OOIs (ablation) | Two-option agent · Modified DuplicatedInput | cumulative reward per episode | Hurt |
| Use randomly sampled option initiation sets | Modified DuplicatedInput agent with learned option policies · Modified DuplicatedInput | cumulative reward per episode | Improved |
| Use LSTM memory over options instead of OOIs | Modified DuplicatedInput agent · Modified DuplicatedInput | cumulative reward per episode | Hurt |
| Reduce the option set from 14 options to 8 or 4 while retaining OOIs (ablation) | TreeMaze agent · TreeMaze | cumulative reward per episode | Hurt |
| Use OOIs rather than standard initiation sets | TreeMaze agent with 14 options · TreeMaze | cumulative reward per episode | Improved |
| Use LSTM memory over options instead of OOIs | TreeMaze agent · TreeMaze | episodes to learn the optimal policy: more than 100K episodes | Hurt |
Closest papers by meaning
- Provable Representation with Efficient Planning for Partially Observable Reinforcement Learning (ICML 2024 · similarity 0.89)
- Rethinking Transformers in Solving POMDPs (ICML 2024 · similarity 0.88)
- Scalable Policy-Based RL Algorithms for POMDPs (NeurIPS 2025 · similarity 0.88)
- Model-based Reinforcement Learning for Confounded POMDPs (ICML 2024 · similarity 0.88)
- Sample-Efficient Learning of POMDPs with Multiple Observations In Hindsight (ICLR 2024 · similarity 0.88)
- Partially Observable Hierarchical Reinforcement Learning with AI Planning (Student Abstract) (AAAI 2024 · similarity 0.88)
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.
