Bayesian model-based agglomerative sequence segmentation

Bayesian algorithm for segmenting real-valued input-output data into non-overlapping segments

The Bayesian model-based agglomerative sequence segmentation (BMASS) algorithm partitions a sequence of real-valued input-output data into non-overlapping segments. The segment boundaries are chosen under the assumption that, within each segment, the data follow a multi-variate linear model.

Segmentation is agglomerative and consists of greedily merging pairs of consecutive segments. Initially, each datum is placed in an individual segment. In each iteration, a single pair of segments is merged based on the log-likelihood ratio of the merge hypothesis. The merging process continues until the log-likelihood ratio becomes negative, or until all segments have been merged.

This submission includes a test function that generates a set of synthetic data and compares the true segment boundaries against those identified by the BMASS algorithm.

If you find this submission useful for your research/work please cite my MathWorks community profile. Feel free to contact me directly if you have any technical or application-related questions.

INSTRUCTIONS:

After downloading this submission, extract the compressed file inside your MatLab working directory and run the test function (bmasstest.m) for a demonstration.