During Fall 2004, the Knowledge Discovery and Data Mining Research Lab (KDDM) at CIS UAB, led by Dr. Alan Sprague and Dr. Chengcui Zhang, was established in order to support research of new algorithms, systems, and applications for large-scale data mining and visualization. The research combines development of pattern matching algorithms, tatistical techniques, distributed database techniques, and visualization methods. It hosts faculty, students, and visiting scholars, conducting cross-disciplinary research as well as developing test beds.
Our current research activities focus on the following areas:
Event sequence data mining
Multimedia data mining, in particular images and videos
Spatio-temporal data mining (e.g., traffic surveillance data)
Meta-learning for model selection and combination
Incremental learning, i.e., adapting to new data without retraining
Distributed data mining for large scale scientific data using grid computing
Data mining for Biomedical Informatics
Computer Forensics (e.g., spam and phishing data mining)
We have applied research to several domains, with close collaboration with cyber-security specialists, colleagues in Physical Medicine & Rehabilitation, Bio-statistics, and Government, as well as with industrial collaborators such as IBM and eBay. The methods and tools have so far been applied to healthcare applications, traffic surveillance applications, image analysis and retrieval (e.g., canonical view extraction, object-based image retrieval, and image spam mining), the identification of events of interest for sports videos, bio-medical image/video mining (e.g., histological image analysis for skin cancer screening), and email spam and phishing kit data mining. More recently, our research areas have been expanded to include social science applications and biomedical text mining. Some highlight systems include analysis of organizational patterns of lobbying activities and automatic extraction of gene co-expression hypotheses from published biomedical literature. Our research is currently supported by NSF and NIH.