Research map / Reinforcement learning / Visual and representation RL
Dynamics-Aware Unsupervised Discovery of Skills
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 DADS skills with latent-space MPPI instead of standard model-based RL | Ant navigation · Ant navigation to goals | normalized average distance to goal (Δ) | Helped |
| Use DADS skills with latent-space MPPI instead of goal-conditioned RL | Ant navigation · Dense-reward Ant navigation | navigation performance | Mixed |
| Use DADS skills with latent-space MPPI instead of goal-conditioned RL | Ant navigation · Sparse-reward Ant navigation | navigation performance | Helped |
| Learn primitives with DADS predictability-and-diversity objective instead of DIAYN diversity objective | Ant navigation skills · Ant x-y trajectory variance | trajectory variance | Helped |
| Learn primitives with DADS instead of DIAYN | Hierarchical Ant navigation · Ant navigation to goals | navigation performance | Helped |
| Use continuous primitives instead of discrete primitives (ablation) | DADS · Ant navigation | planning performance | Helped |
| Add an x-y prior to DADS skill-dynamics (ablation) | DADS Ant skills · Ant x-y trajectories | skill variance | Helped |
| Remove the x-y prior from DADS (ablation) | DADS Ant skills · Ant skill-dynamics prediction | prediction error growth | Hurt |
| Resample the skill at every timestep instead of keeping it temporally extended (ablation) | DADS · Unsupervised skill discovery | skill learning quality | Hurt |
Closest papers by meaning
- Unsupervised Domain Adaptation with Dynamics-Aware Rewards in Reinforcement Learning (NeurIPS 2021 · similarity 0.92)
- Efficient Skill Discovery via Regret-Aware Optimization (ICML 2025 · similarity 0.90)
- Constrained Ensemble Exploration for Unsupervised Skill Discovery (ICML 2024 · similarity 0.90)
- Skill Disentanglement in Reproducing Kernel Hilbert Space (AAAI 2025 · similarity 0.89)
- Robust Policy Learning via Offline Skill Diffusion (AAAI 2024 · similarity 0.89)
- Learning from Reward-Free Offline Data: A Case for Planning with Latent Dynamics Models (NeurIPS 2025 · similarity 0.89)
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.