Enhancing Clustering by Exploiting Complementary Data Modalities in the Medical Domain

• Published in: Artificial Intelligence Applications and Innovations Vol. 381; pp. 357 - 367
• Main Authors: Fodeh, Samah Jamal, Haddad, Ali, Brandt, Cynthia, Schultz, Martin, Krauthammer, Michael
• Format: Book Chapter Journal Article
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 01.09.2012
• Series: IFIP Advances in Information and Communication Technology
• Subjects: BI-NMF; biomedical; clustering; non negative matrix factorization; non negative matrix factorization; images; clustering; BI-NMF; biomedical
• ISBN: 9783642334085; 3642334083
• ISSN: 1868-4238; 1861-2288
• DOI: 10.1007/978-3-642-33409-2_37

Data Clustering has been an active area of research in many different application areas, with existing clustering algorithms mostly focusing on partitioning one modality or representation of the data. In this study, we delineate and demonstrate a new, enhanced data clustering approach whose innovation is its exploitation of multiple data modalities. We propose BI-NMF, a bi-modal clustering approach based on Non Negative Matrix Factorization (NMF) that clusters two differing data modalities simultaneously. The strength of our approach is its combining of multiple aspects of the data when forming the final clusters. To assess the utility of our approach, we performed several experiments on two distinct biomedical datasets with two modalities each. Comparing the clusters of BI-NMF with NMF clusters of single data modality, we observed consistent performance enhancement across both datasets. Our experimental results suggest that BI-NMF is advantageous for boosting data clustering.