Research map / Reinforcement learning / Visual and representation RL
Latent Representation Alignment for Offline Goal-Conditioned Reinforcement Learning
Hyungkyu Kang, Byeongchan Kim, Min-hwan Oh
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 continuity regularization (ablation) | LAVL · pointmaze-giant | success rate (%): 35% -> 95% | Improved |
| add continuity regularization (ablation) | LAVL · antmaze-giant and scene | success rate | Improved |
| vary latent dimension from 16 to 256 (ablation) | LAVL · OGBench | success rate | No clear effect |
| use LAN value architecture instead of IQE, MRN or Hilbert (ablation) | LAVL · scene | success rate: 88% vs below 20% | Improved |
| use LAN instead of MLP value (HIQL baseline) (ablation) | LAVL · scene | success rate: below 40% -> 88% | Improved |
| use IQE, MRN or Hilbert instead of MLP value (ablation) | LAVL · antmaze-giant | success rate | Improved |
| replace HIQL high-level value with LAVL-HV separate high-level value (ablation) | HIQL · two maze datasets | success rate: nearly 0% -> over 80% | Improved |
| use unitary LAVL value instead of separate high/low-level values (LAVL-HV) (ablation) | LAVL · OGBench | success rate | Improved |
| use LAVL instead of HIQL | HIQL · maze giant average | success rate (%): 21 -> 85 | Improved |
| use LAVL instead of OTA on humanoidmaze-giant-navigate | OTA · humanoidmaze giant-navigate | success rate (%): 92 -> 83 | Hurt |
| use LAN instead of MLP in GCXVL (ablation) | GCXVL · Antmaze-large-navigate, beta 3 | success rate (%): 17 -> 54 | Improved |
| use LAN instead of MLP in GCXVL (ablation) | GCXVL · Cube-single-play, beta 1 | success rate (%): 24 -> 5 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- HIQL — HIQL: Offline Goal-Conditioned RL with Latent States as Actions (NeurIPS 2023) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Option-aware Temporally Abstracted Value for Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2025 · similarity 0.92)
- Physics-informed Value Learner for Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2025 · similarity 0.92)
- Conservative Offline Goal-Conditioned Implicit V-Learning (ICML 2025 · similarity 0.92)
- Stitching Sub-trajectories with Conditional Diffusion Model for Goal-Conditioned Offline RL (AAAI 2024 · similarity 0.90)
- OGBench: Benchmarking Offline Goal-Conditioned RL (ICLR 2025 · similarity 0.90)
- Flattening Hierarchies with Policy Bootstrapping (NeurIPS 2025 · similarity 0.90)
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.
