The research conducted in the Knowledge Discovery and Data Mining Research (KDDM) Lab combines development of pattern matching algorithms, statistical techniques, distributed database techniques, and visualization methods.
The current research activities of KDDM focus on the following areas:
- multimedia data mining, in particular images and videos
- event/anomaly detection and analysis
- spatio-temporal data mining
- social network mining
- high performance data mining
- computer forensics
We have applied research to several domains, and collaborate closely with cyber-security specialists, colleagues in the departments of Physical Medicine & Rehabilitation, Biostatistics, and Government, and with industrial collaborators such as IBM and eBay. Our methods and tools have been applied to:
- healthcare applications
- traffic surveillance applications
- image analysis and retrieval (canonical view extraction, object-based image retrieval, and image spam mining)
- identification of events of interest for sports videos
- bio-medical image/video mining (histological image analysis for skin cancer screening)
- email spam and phishing kit data mining
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.