Scalable Node-Level Computation Kernels for Parallel Exact Inference

In this paper, we investigate data parallelism in exact inference with respect to arbitrary junction trees. Exact inference is a key problem in exploring probabilistic graphical models, where the computation complexity increases dramatically with clique width and the number of states of random varia...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE transactions on computers Ročník 59; číslo 1; s. 103 - 115
Hlavní autoři: Yinglong Xia, Prasanna, V.K.
Médium: Journal Article
Jazyk:angličtina
Vydáno: IEEE 01.01.2010
Témata:
Bayesian network
Computational modeling
Exact inference
Inference algorithms
junction tree
Junctions
message passing
node-level primitives
Particle separators
Program processors
Random variables
ISSN:0018-9340
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In this paper, we investigate data parallelism in exact inference with respect to arbitrary junction trees. Exact inference is a key problem in exploring probabilistic graphical models, where the computation complexity increases dramatically with clique width and the number of states of random variables. We study potential table representation and scalable algorithms for node-level primitives. Based on such node-level primitives, we propose computation kernels for evidence collection and evidence distribution. A data parallel algorithm for exact inference is presented using the proposed computation kernels. We analyze the scalability of node-level primitives, computation kernels, and the exact inference algorithm using the coarse-grained multicomputer (CGM) model. According to the analysis, we achieve O(Nd c w c Pi j=1 wc r C,j /P) local computation time and O(N) global communication rounds using P processors, 1 les P les max c PiPi j1 wc r C,j, where N is the number of cliques in the junction tree; d c is the clique degree; r C,j is the number of states of the jth random variable in C; wc is the clique width; and w s is the separator width. We implemented the proposed algorithm on state-of-the-art clusters. Experimental results show that the proposed algorithm exhibits almost linear scalability over a wide range.
ISSN:0018-9340
DOI:10.1109/TC.2009.106