AIMar 27, 2013
A Constraint Propagation Approach to Probabilistic Reasoning
arXiv:1304.3422v1226 citations
Originality Synthesis-oriented
AI Analysis
This work addresses uncertainty management in AI systems, but it appears incremental as it builds on existing probabilistic and constraint-based methods.
The paper tackled the challenge of integrating constraint propagation with probabilistic reasoning, showing that both predictive and diagnostic inferences can operate concurrently and converge to a stable equilibrium.
The paper demonstrates that strict adherence to probability theory does not preclude the use of concurrent, self-activated constraint-propagation mechanisms for managing uncertainty. Maintaining local records of sources-of-belief allows both predictive and diagnostic inferences to be activated simultaneously and propagate harmoniously towards a stable equilibrium.