Name of the Speaker: Shivani Bathla (EE17D037)
Guide: Dr.Nitin chandrachoodan
Venue/Online meeting link: ESB 210B
Date/Time: 20th September 2022,10AM
Exact inference in Bayesian networks is intractable and has an exponential dependence on the size of the largest clique in the corresponding clique tree (CT), necessitating approximations. Factor based methods to bound clique sizes are more accurate than structure based methods, but are expensive since they involve inference of beliefs in a large number of candidate structure or region graphs. We propose an alternative approach for approximate inference based on an incremental build-infer-approximate (IBIA) paradigm, which converts the Bayesian network into a data structure containing a sequence of linked clique tree forests (SLCTF), with clique sizes bounded by a user-specified value. We show that our algorithm for incremental construction of clique trees always generates a valid CT and our approximation technique preserves the joint beliefs of the variables within a clique.In this seminar, I will present results obtained using the proposed algorithm for incremental construction of CTs and inference of prior beliefs using the IBIA framework. The framework was used to evaluate signal and transition probabilities for large digital circuits as well as several other Bayesian network benchmarks. The results show a significant reduction in error when compared to other approximate methods with competitive runtimes.