Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Learning Human Habits with Rule-Guided Active Inference
GONG ZHIREN, Chao Yang, Wendi Ren, Shuang Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 the habitual rule bank (ablation) | rule-guided active inference · NBA SportVU | Acc@3: 79.5 -> 91.4 | Improved |
| add the habitual rule bank (ablation) | rule-guided active inference · NBA SportVU | Latency (ms): 73 -> 36 | Improved |
| add the latent intention state m_t (ablation) | rule-guided active inference · Atari-Berzerk | Acc@3: 70.1 -> 77.3 | Improved |
| add VFE generative consistency to the objective (ablation) | rule-guided active inference · Car-Following | Acc@3: 78.5 -> 95.9 | Improved |
| use consolidated rule selection instead of greedy rule selection (ablation) | rule-guided active inference · NBA SportVU, Car-Following, DDXPlus, Atari | Acc@k and HHAR | Mixed |
| grow rule bank beyond the Pareto knee (ablation) | rule-guided active inference · NBA SportVU | Acc@k: optimal at RC=6; more rules degrade | Hurt |
| grow rule bank from none to compact size (ablation) | rule-guided active inference · NBA SportVU | Latency and Acc@3 | Improved |
| use rule-guided active inference instead of DreamerV2 | NBA SportVU · NBA SportVU | Acc@3: 83.57 -> 91.32 | Improved |
| use rule-guided active inference instead of DAI-MC | DDXPlus (URTI) · DDXPlus (URTI) | Latency (ms): 2304.23 -> 159.45 | Improved |
| use rule-guided active inference instead of Qwen-0.5B | DDXPlus (URTI) · DDXPlus (URTI) | Acc@3: 19.62 -> 73.58 | Improved |
Closest papers by meaning
- Evolving Minds: Logic-Informed Inference from Temporal Action Patterns (ICML 2025 · similarity 0.87)
- Habitizing Diffusion Planning for Efficient and Effective Decision Making (ICML 2025 · similarity 0.85)
- Reason to Play: Behavioral and Brain Alignment Between Frontier LRMs and Human Game Learners (NeurIPS 2026 · similarity 0.85)
- Beyond Single Stationary Policies: Meta-Task Players as Naturally Superior Collaborators (NeurIPS 2024 · similarity 0.85)
- Behavioral Exploration: Learning to Explore via In-Context Adaptation (ICML 2025 · similarity 0.85)
- GUIDE: Real-Time Human-Shaped Agents (NeurIPS 2024 · similarity 0.84)
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.
