首页> 外文会议>International Conference on Soft Computing for Problem Solving >Frequent Item Set Mining of Large Datasets Using CUDA Computing
【24h】

Frequent Item Set Mining of Large Datasets Using CUDA Computing

机译:使用CUDA Computing频繁的项目设置大型数据集的挖掘

获取原文

摘要

Frequent item set mining is a very popular method in data mining and is used extensively to find out the most recurring items in mainly market basket analysis. It is commonly used for association rule learning. The market basket analysis is seen to be used in many fields like Web mining and intrusion detection. There are many algorithms like Apriori algorithm, etc., to find out the frequent items in given data but these are more of a sequential approach and take huge amounts of time for large sets of data. Existing Soft computing-based approaches for solving frequent item set mining like genetic algorithm and fuzzy logic systems are proven to reduce execution time but not in a scale in which a massively parallel system does. So the main objective of this paper is to accelerate frequent item set mining process using GPU's CUDA architecture. We have also performed a comparative study with parallel version of frequent item set mining using openMP. Our results show speedup of 2.2 for CUDA over serial implemented using genetic algorithm and 1.8 for CUDA over OpenMP.
机译:频繁项集挖掘是数据挖掘中一个非常流行的方法,并广泛使用,以找出主要市场购物篮分析的最经常性项目。它通常用于关联规则的学习。市场购物篮分析被认为是在像Web挖掘和入侵检测许多领域使用。还有像Apriori算法等多种算法,找出在给定的数据经常项目,但这些都是比较顺序的做法,并采取大量的时间用于大型数据集。现有的解决频繁项集挖掘像遗传算法和模糊逻辑系统中的软基于计算的方法已被证明可以减少执行时间,但不会在其中大规模并行系统做了规模。因此,本文的主要目的是加速使用GPU的CUDA架构频繁项集挖掘过程。我们还进行与使用OpenMP频繁项集矿山开采的水货版本的比较研究。我们的研究结果表明2.2加速的CUDA比使用遗传算法和1.8 CUDA上的OpenMP实现串行。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号