Research map / Reinforcement learning / Visual and representation RL
DeepSynth: Automata Synthesis for Automatic Task Segmentation in Deep 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 |
|---|---|---|---|
| Use automata-guided DeepSynth instead of DQN | Minecraft tasks · minecraft-t1 | episodes to convergence: 40 -> 25 | Helped |
| Use automata-guided DeepSynth instead of DQN | Minecraft tasks · minecraft-t3 | episodes to convergence: t/o -> 40 | Helped |
| Use automata-guided DeepSynth instead of DQN | Minecraft tasks · minecraft-t7 | episodes to convergence: t/o -> 70 | Helped |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · robot-surve | episodes to convergence: 10 -> 10 | No clear effect |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · slp-easy-sml | episodes to convergence: 10 -> 10 | No clear effect |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · slp-easy-med | episodes to convergence: 20 -> 20 | No clear effect |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · slp-easy-lrg | episodes to convergence: 30 -> 30 | No clear effect |
| Use automata-guided DeepSynth instead of DQN | SLP hard tasks · slp-hard-sml | episodes to convergence: t/o -> 80 | Helped |
| Use automata-guided DeepSynth instead of DQN | SLP hard tasks · slp-hard-med | episodes to convergence: t/o -> 100 | Helped |
| Use automata-guided DeepSynth instead of DQN | SLP hard tasks · slp-hard-lrg | episodes to convergence: t/o -> 120 | Helped |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · frozen-lake-1 | episodes to convergence: 120 -> 100 | Helped |
| Use automata-guided DeepSynth instead of DQN | DeepSynth and DQN · frozen-lake-2 | episodes to convergence: 150 -> 150 | No clear effect |
Closest papers by meaning
- Induction of Subgoal Automata for Reinforcement Learning (AAAI 2020 · similarity 0.86)
- HDCS: Hierarchy Discovery and Critic Shaping for Reinforcement Learning with Automaton Specification (NeurIPS 2026 · similarity 0.86)
- ARM-FM: Automated Reward Machines via Foundation Models for Compositional Reinforcement Learning (ICLR 2026 · similarity 0.86)
- Integrating Planning and Deep Reinforcement Learning via Automatic Induction of Task Substructures (ICLR 2024 · similarity 0.86)
- DeepMath-103K: A Large-Scale, Challenging, Decontaminated, and Verifiable Mathematical Dataset for Advancing Reasoning (ICLR 2026 · similarity 0.85)
- DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search (ICLR 2026 · 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.