Learning Globally Optimal Dynamic Bayesian Network with the Mutual Information Test (MIT) Criterion

Dynamic Bayesian networks (DBN) are widely applied in modeling various biological networks, including the gene regulatory network. Due to several NP-hardness results on learning static Bayesian network, most methods for learning DBN are heuristic, that employ either local search such as greedy hill-climbing, or a meta optimization framework such as genetic algorithm or simulated annealing.

We present GlobalMIT, a toolbox for learning the globally optimal DBN structure using a recently introduced information theoretic based scoring metric named mutual information test (MIT). Under MIT, learning the globally optimal DBN can be efficiently achieved in polynomial time. The toolbox is implemented in Matlab, with also a C++ stand-alone implementation of the search engine for improved performance.

This research work was part of a larger project (BF040037) funded by Australia-India Strategic Research Fund. The chief investigators were A/Prof. Madhu Chetty(Federation University), Prof. Ross Coppel (Monash) and Prof. Pramod Wangikar (IIT Bombay). It was carried out at Monash University (Gippsland) which is now a part of Federation University Australia.

This site is managed by Vinh Nguyen. Inquiries and Feedbacks are welcome at vinh.nguyen at unimelb.edu.au or vinh.nguyenx at gmail.com.

If you find our work useful for you, please cite:

Vinh, N. X., Chetty, M., Coppel, R., and Wangikar, P. P. (2011). GlobalMIT: Learning Globally Optimal Dynamic Bayesian Network with the Mutual Information Test (MIT) Criterion, Bioinformatics, doi: 10.1093/bioinformatics/btr457, Pre-publication PDF Publisher's site PDF