The amount of data relevant to home land defense is massive, distributed and growing rapidly through the addition of high volume data streams and feeds. This presents fundamentally new mathematical challenges. These relate to: 1) the real time and near real time detection of significant events in high volume data streams; 2) the forensic analysis of massive amounts of archived data to uncover patterns and events of interest; and 3) the mining of distributed data, which for a variety of reasons will never be centrally warehoused. To complicate matters further, homeland defense must concern itself with a variety of different data types, including, signals, text, images, transaction data, streaming media, web data, and computer to computer traffic.
The event will bring together researchers from a variety of fields for tutorials and specialized talks about these challenges. The tutorial, which runs from Monday to Wednesday, will present to non-experts or those wanting a coherent introduction to the field a variety of tools that are relevant to the topics described. The workshop, which runs from Thursday through Saturday, will contain more specialized talks. It is possible to register for the tutorial alone, the workshop alone, or both.
There will be tutorials on text mining, parallel data mining, algorithmic issues in processing data streams, database support for data mining, on-line learning, forensic ring analysis, and data fusion, as well as a number of survey talks.
The workshop will include talks on algorithmic issues in processing streaming data, text mining & classification, anomaly detection, outlier analysis, forensic data analysis, on-line learning, real-time data mining, parallel data mining, visualization and data mining, and mining graphical data.