Hauptnavigation

Bibkey Proceedings of the RapidMiner Community Meeting And Conference (RCOMM 2010)
Author Fricke, Peter and Stolpe, Marco
Ls8autor Fricke, Peter
Stolpe, Marco
Editor http://www.rcomm2010.com
Booktitle Proceedings of the RapidMiner Community Meeting And Conference (RCOMM 2010)
Institution Huge masses of data and potentially infinite data streams pose big challenges to methods in data mining that analyse data off-line and in several passes. In the area of intrusion detection, algorithms that detect characteristical patterns in system call data could have to process several hundred megabytes of data per minute. We describe a plugin that contains algorithms for the aggregation of data streams by determining frequent tuples of hierarchical elements, so called Hierarchical Heavy Hitters. We further discuss more general questions concerning the current state of RapidMiner in relation to stream mining.
Abstract Huge masses of data and potentially infinite data streams pose big challenges to methods in data mining that analyse data off-line and in several passes. In the area of intrusion detection, algorithms that detect characteristical patterns in system call data could have to process several hundred megabytes of data per minute. We describe a plugin that contains algorithms for the aggregation of data streams by determining frequent tuples of hierarchical elements, so called Hierarchical Heavy Hitters. We further discuss more general questions concerning the current state of RapidMiner in relation to stream mining.
Year 2010
Url http://www.rcomm2010.com



  • Impressum