Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Relative Variational Intrinsic Control
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 |
|---|---|---|---|
| Add second predictor conditioned only on final state (two-predictor relative reward) instead of VIC (ablation) | VIC skill learning · DeepMind Control Suite Reacher | skill dependence on start state | Helped |
| Use RVIC rather than VIC skills | skill learning · Seaquest and Montezuma's Revenge | skill generalization across start positions | Helped |
| Use RVIC skills plus primitive actions in R2D2 meta-controller | R2D2 HRL · six Atari games | learning speed over 40M learner steps | Mixed |
| Use dense skill reward at every timestep instead of sparse final-step reward (ablation) | VIC and RVIC · skill learning | empirical skill quality | Helped |
| Zero discount at skill end (ablation) | skill learning · Atari | performance | Helped |
| Fixed discount with bootstrapping between skill episodes (ablation) | skill learning · DeepMind Control Suite | performance | Helped |
| Add meta-action cost (0.0, 0.1, 0.15) to the meta-controller (ablation) | HRL meta-controller · six Atari games | HRL return: 0.0, 0.1, 0.15 | Helped |
| Vary skill episode length (10, 15, 25) (ablation) | HRL skills · six Atari games | HRL return: 10, 15, 25 | Mixed |
Closest papers by meaning
- Skill-aware Mutual Information Optimisation for Zero-shot Generalisation in Reinforcement Learning (NeurIPS 2024 · similarity 0.87)
- When Do Skills Help Reinforcement Learning? A Theoretical Analysis of Temporal Abstractions (ICML 2024 · similarity 0.87)
- Unifying Goal-Conditioned RL and Unsupervised Skill Learning via Control-Maximization (NeurIPS 2026 · similarity 0.87)
- Controlling Behavioral Diversity in Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.87)
- Policy Compatible Skill Incremental Learning via Lazy Learning Interface (NeurIPS 2025 · similarity 0.86)
- Exploration by Learning Diverse Skills through Successor State Representations (NeurIPS 2024 · similarity 0.86)
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.