We consider a CPU constrained environment for finding approximation of frequent sets in data streams using the landmark window. Our algorithm can detect overload situations, i.e., breaching the CPU capacity, and sheds data in the stream to “keep up”. This is done within a controlled error threshold by exploiting the Chernoff-bound. Empirical evaluation of the algorithm confirms the feasibility.
Lecture Notes in Computer Science, 2012, Vol 7520, p. 590-597
Main Research Area:
International Conference on Scalable Uncertainty Management, 2012