|International Journal of Computer Applications
|Foundation of Computer Science (FCS), NY, USA
|Volume 105 - Number 7
|Year of Publication: 2014
|Authors: Dharmesh Bhalodia, Chhaya Patel
Dharmesh Bhalodia, Chhaya Patel . Mining Frequent Patterns with Optimized Candidate Representation on Graphics Processor. International Journal of Computer Applications. 105, 7 ( November 2014), 1-8. DOI=10.5120/18386-9634
Frequent itemset mining algorithms mine subsets of items that appear frequently in a collection of sets. FIM is a key investigation in numerous data mining applications, and the FIM tools are among the most computationally demanding in data mining. In this research paper we present a new approach to represent candidate in parallel Frequent Itemset Mining algorithm. Our new approach is extension of GPApriori, a GP-GPU version of FIM. This implementation is optimized to achieve high performance on a heterogeneous platform consisting of a shared memory multiprocessor and multiple cores NVIDIA based Graphics Processing Unit (GPU) coprocessor. An experiments compared with the GPApriori on NVIDIA Kepler GPUs and observed 1. 5X to 2X required less memory and significant improvements in time relative to GPApriori.