Efficient Search-Based Inference for Noisy-OR Belief Networks: TopEpsilon
This addresses the challenge of scalable inference in probabilistic graphical models for researchers and practitioners, but it appears incremental as it builds on prior specialized methods.
They tackled the problem of inefficient inference for large, complex belief networks by developing a search-based algorithm for approximate inference on arbitrary noisy-OR belief networks, generalizing earlier work, with initial experimental results showing promise.
Inference algorithms for arbitrary belief networks are impractical for large, complex belief networks. Inference algorithms for specialized classes of belief networks have been shown to be more efficient. In this paper, we present a search-based algorithm for approximate inference on arbitrary, noisy-OR belief networks, generalizing earlier work on search-based inference for two-level, noisy-OR belief networks. Initial experimental results appear promising.