Research map / Multi-agent systems and games / Human-AI and goal-conditioned
A Boolean Task Algebra for Reinforcement Learning
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 |
|---|---|---|---|
| Learn extended value functions instead of regular value functions | Q-learning on Four Rooms · Four Rooms (sample cost) | cumulative samples to learn base tasks | Hurt |
| Use extended value functions with Boolean composition instead of disjunction-only composition | Zero-shot composition · Four Rooms | number of tasks solvable without further learning: 2^(2^K) tasks vs 2^K - 1 | Helped |
| Use Boolean composition on 40-goal Four Rooms instead of disjunctive approach | Zero-shot composition · Four Rooms, 40 goals | base tasks required: 40 -> 7 | Helped |
| Compose extended UVFAs (OR, AND, XOR) of Blue and Square tasks | Deep Q-learning with UVFA · Object-collection video game (84x84 RGB) | average return | Helped |
Closest papers by meaning
- Generalisation in Lifelong Reinforcement Learning through Logical Composition (ICLR 2022 · similarity 0.90)
- Composing Value Functions in Reinforcement Learning (ICML 2019 · similarity 0.89)
- Unifying Task Specification in Reinforcement Learning (ICML 2017 · similarity 0.87)
- Run-Time Task Composition with Safety Semantics (ICML 2024 · similarity 0.86)
- Skill Machines: Temporal Logic Skill Composition in Reinforcement Learning (ICLR 2024 · similarity 0.86)
- PUZZLES: A Benchmark for Neural Algorithmic Reasoning (NeurIPS 2024 · 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.