From Constraints to Cognition: A Hybrid Framework for Adaptive, Explainable(…)
summary
A hybrid pipeline compiles grounded Answer Set Programming schedules into an epistemic layer (L-DINF) via a named tau translation that turns baseline atoms into beliefs, intentions, feasibility checks and preferences; hard ASP constraints remain active during repairs. An executable DALI prototype on SICStus Prolog runs a healthcare scenario, performs bounded local repair and inter-group lending, logs timestamped explanations, restores feasibility, and reports zero hard-constraint violations while keeping a fallback to ASP re-optimization for non-local failures.
AI-written summary · about
chapters (25)
- 0:00Hybrid scheduling framework intro
- 1:05Gap in static scheduling systems
- 2:08Hybrid ASP and LDIMP architecture
- 3:11Technical contributions
- 6:01Healthcare constraints and disruptions
- 7:25Patient archetype instantiation
- 8:28LDIMP fragment semantics
- 10:38Agent roles and entities
- 12:05Compilation function tau
- 13:49Baseline appointment example
- 14:52Constraint preservation in repair
- 15:55Two-clinic scenario example
- 16:58Simulating disruption and repair
- 18:02Epistemic model transitions
- 19:07Why Dali for adaptation
- 20:11Dali prototype implementation
- 21:57Executable constructs realization
- 23:00Dali code implementation
- 24:04Delegation communication protocol
- 25:28Execution trace
- 26:34Explanations from rule logs
- 26:55Results validation
- 27:16Future research directions
- 28:00Conclusion
- 28:21Q&A on deontic logic
related talks
50mFrom Prompt Hacking to Architectural Determinism: Engineering Reliable GenAI Systems
31mFormalizing and Reasoning about Contracts between Agents
10mFrom Philosophy to Interfaces: an Explanatory Method and a Tool Inspired by Achinstein's Theory ...
22mA Cognitive Framework for Self-Adjusting Systems Through Experience Feedback.
56mPeople, Ideas, and the Path Ahead
21mLazy Validation and Self-Healing for Agentic Programs
39mMaking AI Reliable in the SDLC: The Role of Deterministic Context at Scale
20mBuilding Deterministic Infrastructure for Non Deterministic AI Agents