A1294
Title: Solution diversification in graph matching matched filters
Authors: Zhirui Li - University of Pennsylvania (United States) [presenting]
Ben Johnson - Jataware (United States)
Daniel Sussman - Boston University (United States)
Carey Priebe - Johns Hopkins University (United States)
Vince Lyzinski - University of Maryland, College Park (United States)
Abstract: The purpose is to present a novel approach for finding multiple noisily embedded template graphs in a very large background graph. The method builds upon the graph-matching-matched-filter technique proposed in a prior study, with the discovery of multiple diverse matchings being achieved by iteratively penalizing a suitable node-pair similarity matrix in the matched filter algorithm. In addition, algorithmic speed-ups are proposed that greatly enhance the scalability of the matched-filter approach. Theoretical justification of the methodology is presented in the setting of correlated Erdos-Renyi graphs, showing its ability to sequentially discover multiple templates under mild model conditions. The method's utility is additionally demonstrated via extensive experiments both using simulated models and real-world datasets, including human brain connectomes and a large transactional knowledge base.
