<< Click to Display Table of Contents >> Navigation:  Building blocks of QGeNIe > Inference algorithms > Relevance-based decomposition

Relevance-based decomposition is an exact algorithm based on the clustering algorithm that performs a decomposition of the network when the network is very large. The algorithm was described in (Lin & Druzdzel, 1997). Relevance-based decomposition extends the boundary of what is computable, while gracefully changing into the clustering algorithm for small networks. Because there is some overhead related to decomposition, we suggest that this algorithm be used only when the clustering algorithm cannot handle your networks.