Parallelizing Coherent Rule Mining Algorithm on CUDA

Aditya A. Davale, Shailendra W. Shende

Abstract: In data mining, association rules are discovered from the domain when the minimum support threshold value is given by domain expert. Minimum support threshold affects the number of rules generated and the quality of rules. So In this work, a framework is proposed to discover domain knowledge in terms of coherent rules. This rules are discovered from properties of propositional logic, and does not require minimum support threshold. The execution of association rule mining algorithm involves processing of huge data. Handling such amount of data having millions of rows is very challenging. High-end computers and server-side machines are used to process such algorithms for enormous amount of data but are very expensive and accessible to only a few. In comparison graphics processing units (GPUs) have much high computation power and are also less expensive. So, GPU acts as the high performance co-processors. So, the implementation of association rule mining algorithm on GPU will provides the high performance computing and increase in the speedup.

Keywords: Data mining, GPU, CUDA, parallel processing

How to Cite?: Aditya A. Davale, Shailendra W. Shende, "Parallelizing Coherent Rule Mining Algorithm on CUDA", Volume 4 Issue 4, April 2015, International Journal of Science and Research (IJSR), Pages: 1076-1079, https://www.ijsr.net/getabstract.php?paperid=SUB153118, DOI: https://dx.doi.org/10.21275/SUB153118

Download Citation: APA | MLA | BibTeX | EndNote | RefMan