Research map / Reinforcement learning / Visual and representation RL
Deep Surrogate Assisted Generation of Environments
Varun Bhatt, Bryon Tjanaka, Matthew Christopher Fontaine, Stefanos Nikolaidis
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 ancillary (occupancy grid) prediction and downsampling to basic surrogate (ablation) | DSAGE Basic · Maze | QD-score: 11,740.00 -> 16,446.60 | Improved |
| add ancillary (occupancy grid) prediction and downsampling to basic surrogate (ablation) | DSAGE Basic · Mario | QD-score: 1,306.11 -> 4,362.29 | Improved |
| add ancillary agent behavior data prediction (ablation) | DSAGE-Only Down · Maze | QD-score: 14,205.20 -> 16,446.60 | Improved |
| add ancillary agent behavior data prediction (ablation) | DSAGE-Only Down · Mario | QD-score: 4,067.42 -> 4,362.29 | No clear effect |
| add ancillary data prediction (ablation) | DSAGE Basic · Maze | MAE of mean agent path length: 157.69 -> 96.58 | Improved |
| add ancillary data prediction (ablation) | DSAGE-Only Down · Mario | MAE of number of jumps: 6.52 -> 7.16 | Hurt |
| add downsampling of surrogate archive (ablation) | DSAGE-Only Anc · Maze | QD-score: 14,568.00 -> 16,446.60 | Improved |
| add downsampling of surrogate archive (ablation) | DSAGE-Only Anc · Mario | QD-score: 2,045.28 -> 4,362.29 | Improved |
| add downsampling (ablation) | DSAGE-Only Anc · Maze | evaluations to reach target QD-score: 51,919.60 -> 33,930.40 | Improved |
| add downsampling (ablation) | DSAGE-Only Anc · Mario | evaluations to reach target QD-score: 7,727.40 -> 2,464.40 | Improved |
| train longer to match epochs of downsampling (ablation) | DSAGE Basic · Maze | QD-score: 11,740.00 -> 12,618.20 | Improved |
| train longer to match epochs of downsampling (ablation) | DSAGE-Only Anc · Mario | QD-score: 2,045.28 -> 1,679.55 | Hurt |
Closest papers by meaning
- Quality Diversity through Human Feedback: Towards Open-Ended Diversity-Driven Optimization (ICML 2024 · similarity 0.86)
- Proximal Policy Gradient Arborescence for Quality Diversity Reinforcement Learning (ICLR 2024 · similarity 0.86)
- Enabling Adaptive Agent Training in Open-Ended Simulators by Targeting Diversity (NeurIPS 2024 · similarity 0.85)
- DRED: Zero-Shot Transfer in Reinforcement Learning via Data-Regularised Environment Design (ICML 2024 · similarity 0.85)
- Marginal Benefit Driven RL Teacher for Unsupervised Environment Design (AAAI 2025 · similarity 0.85)
- Improving Regret Approximation for Unsupervised Dynamic Environment Generation (NeurIPS 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.
