Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Modeling Probabilistic Commitments for Maintenance Is Inherently Harder than for Achievement
Qi Zhang, Edmund Durfee, Satinder Singh
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 |
|---|---|---|---|
| Use minimal enable duration influence to approximate an achievement commitment | Recipient planning in 1D Walk · 1D Walk (L=14, L0=5, H=15) | value of policy in true model: 10 -> 6 | Hurt |
| Use minimal enable duration influence to approximate a maintenance commitment | Recipient planning in 1D Walk · 1D Walk (Tm=7, pm=0) | value of policy in true model: 6 -> 0 | Hurt |
| Swap minimal enable duration influence for more sophisticated heuristics (minimal value, minimax regret) for achievement commitments | 1D Walk recipient · 1D Walk, Ta in {5,7,10} | suboptimality | No clear effect |
| Use maximal enable duration influence instead of minimal for achievement commitments | 1D Walk recipient · 1D Walk, Ta in {5,7,10} | suboptimality | Hurt |
| Use minimal value or minimax regret timing instead of minimal/maximal enable duration for maintenance commitments | 1D Walk recipient · 1D Walk, Tm in {5,7,10} | suboptimality | Mixed |
| Use minimal enable duration influence for achievement commitments in Gate Control | Gate Control recipient · Gate Control, Ta in {4,6} | suboptimality | No clear effect |
| Add multi-timepoints influence (timesteps 1, floor(T/2), T) for maintenance commitments | Gate Control recipient · Gate Control, Tm in {4,6} | suboptimality | Improved |
| Add multi-timepoints influence for achievement commitments | Gate Control recipient · Gate Control, Ta in {4,6} | suboptimality | No clear effect |
| Lower commitment probability to pm=0 (no guarantee) for maintenance | Minimal enable duration influence · 1D Walk Theorem 2 example | suboptimality | Hurt |
Closest papers by meaning
- Contracting with a Learning Agent (NeurIPS 2024 · similarity 0.83)
- Towards a Science of AI Agent Reliability (ICML 2026 · similarity 0.82)
- ContractBench: Can LLM Agents Preserve Observation Contracts? (NeurIPS 2026 · similarity 0.82)
- Who&When Pro: Can LLMs Really Attribute Failures in AI Agents? (NeurIPS 2026 · similarity 0.82)
- Emergent Risk Awareness in Rational Agents under Resource Constraints (NeurIPS 2025 · similarity 0.82)
- Learning Optimal Contracts: How to Exploit Small Action Spaces (ICLR 2024 · similarity 0.82)
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.
