Research map / Reinforcement learning / Imitation learning
Programmatic Reward Design by Example
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 PRDBE-learned programmatic rewards instead of the default reward | PPO and AGAC · DoorKey-8x8 | frames to reach average return 0.8 | Helped |
| Use PRDBE-learned programmatic rewards instead of default rewards | PPO and AGAC · DoorKey-16x16 | frames to achieve high performance | Helped |
| Use PRDBE-learned programmatic rewards instead of the default reward | PPO and AGAC · KeyCorridor | frames to achieve high performance | Helped |
| Train with the learned program from a smaller environment | AGAC · KeyCorridorS4R3 and KeyCorridorS6R3 | frames to achieve high performance | Helped |
| Use PRDBE-learned programmatic rewards instead of the default reward | PPO · ObstructedMaze-2Dhlb | frames to achieve high performance | Helped |
| Train in a larger environment with the program learned in a smaller environment | AGAC · ObstructedMaze-Full | average return | Helped |
| Reduce expert demonstrations from 10 to 1 (ablation) | PRDBE with PPO or AGAC · DoorKey | frames to reach average return 0.8 | No clear effect |
| Reduce expert demonstrations from 10 to 1 (ablation) | PRDBE · KeyCorridor | performance | No clear effect |
| Reduce expert demonstrations from 10 to 6 (ablation) | PRDBE · ObstructedMaze-2Dhlb | performance | No clear effect |
| Reduce expert demonstrations below 6 (ablation) | PRDBE · ObstructedMaze-2Dhlb | performance | Hurt |
| Replace learned hole assignments with random assignments satisfying the symbolic constraints (ablation) | Programmatic rewards trained with PPO · DoorKey-16x16 | frames to achieve high performance | Hurt |
| Replace learned hole assignments with random assignments satisfying the symbolic constraints (ablation) | Programmatic rewards trained with AGAC · KeyCorridorS4R4 | frames to achieve high performance | Hurt |
Closest papers by meaning
- R*: Efficient Reward Design via Reward Structure Evolution and Parameter Alignment Optimization with Large Language Models (ICML 2025 · similarity 0.89)
- To the Max: Reinventing Reward in Reinforcement Learning (ICML 2024 · similarity 0.88)
- A Unified Linear Programming Framework for Offline Reward Learning from Human Demonstrations and Feedback (ICML 2024 · similarity 0.87)
- Text2Reward: Reward Shaping with Language Models for Reinforcement Learning (ICLR 2024 · similarity 0.87)
- Progress Reward Model for Reinforcement Learning via Large Language Models (NeurIPS 2025 · similarity 0.87)
- Efficient Language-instructed Skill Acquisition via Reward-Policy Co-Evolution (AAAI 2025 · similarity 0.87)
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.