Title: HMMConverter: A toolbox for hidden Markov models with two novel memory efficient parameter training algorithms
Speaker: Philip Lam
Department of Computer Science, University of British Columbia

Hidden Markov models (HMMs) are flexible and powerful statistical tools for biological sequence analysis, many different HMMs have been developed for different applications recently. However, there are several challenges to design and implement an HMM in practice: (1) the time and memory requirements for predictions and parameter training would be heavy deal to long input sequences (2) it is difficult to set up the parameters of the HMMs. (3) it is time consuming and error-prone to implement HMMs using efficient algorithms.

We here propose an HMM-generating tool HMMConverter which help users to set up an HMM and use it for data analysis easily. HMMConverter takes an XML input file for defining an HMM, it provides several widely used HMM sequence decoding and parameter training algorithms, including two novel memory efficient parameter training algorithms. Furthermore, this tool also supports many special features that are especially useful for Bioinformatics applications.