Research map / Reinforcement learning / Visual and representation RL
Bottleneck-Guided Spectral Subgoals For Offline Goal-Conditioned RL
Hebin Liang, Yi Ma, Chenjun Xiao, Zibin Dong, Zilin Cao, Fei Ni, Yifu Yuan, Jianye HAO
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 |
|---|---|---|---|
| replace HIQL-style time-offset subgoals with bottleneck keypoint subgoals (ablation) | BASS · antmaze-large-play-v2 | success rate (%): 83.3 -> 96.0 (offset 25); 18.7 -> 96.0 (offset 5) | Improved |
| replace HIQL-style time-offset subgoals with bottleneck keypoint subgoals (ablation) | BASS · antmaze-large-diverse-v2 | success rate (%): 84.0 -> 98.7 (offset 25); 22.7 -> 98.7 (offset 5) | Improved |
| use very small number of clusters K in spectral clustering (ablation) | BASS · antmaze-large-play, antmaze-giant-stitch | success rate (%): antmaze-large-play 32.0 / 15.3 at small K vs 96.0 at K=20 | Hurt |
| vary number of clusters K over a broad mid range (ablation) | BASS · antmaze-large-play, pointmaze-large-stitch | success rate (%) | No clear effect |
| use very large K (ablation) | BASS · a few environments | success rate (%) | Hurt |
| swap in the keypoint graph from PointMaze-Stitch for AntMaze-Stitch (ablation) | BASS · Ant-Stitch | success rate (%): 81.3 -> 83.3 | Improved |
| swap keypoint graphs across domains on the same map (ablation) | BASS · Ant-Explore | success rate (%): 72.7 (native) vs 87.3 (PointMaze KPs), 65.3 (AntMaze-Stitch KPs) | Mixed |
| transfer frozen low-level diffuser controller trained on AntMaze-Large-Play to other map scales (ablation) | BASS · AntMaze Umaze, Medium, Ultra | success rate (%): 96.0 (Large) vs 98.7, 96.7, 96.7 | No clear effect |
| use BASS instead of HIQL | HIQL · OGBench pointmaze-large-stitch-v0 | average evaluation steps: 905.04 -> 265.33 | Improved |
| use BASS instead of baselines on pointmaze-teleport-navigate | OGBench baselines · OGBench pointmaze-teleport-navigate-v0 | success rate (%): GCIVL 45 vs BASS 22.0 | No clear effect |
| use BASS instead of CRL on visual antmaze navigate | OGBench baselines · OGBench visual-antmaze-large-navigate-v0 | success rate (%): CRL 84 -> BASS 65.3 | Hurt |
| use BASS on long-horizon stitch tasks versus HIQL | HIQL · OGBench antmaze-giant-stitch-v0 | success rate (%): 2 -> 71.3 | Improved |
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
- Improving Offline RL by Blending Heuristics (ICLR 2024 · similarity 0.86)
- Stitching Sub-trajectories with Conditional Diffusion Model for Goal-Conditioned Offline RL (AAAI 2024 · similarity 0.86)
- Horizon Reduction Makes RL Scalable (NeurIPS 2025 · similarity 0.86)
- Are Expressive Models Truly Necessary for Offline RL? (AAAI 2025 · similarity 0.86)
- Diffusion Subgoal Planning for Long-Horizon Offline Goal-Conditioned Reinforcement Learning (NeurIPS 2026 · similarity 0.85)
- OGBench: Benchmarking Offline Goal-Conditioned RL (ICLR 2025 · similarity 0.85)
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.
