Release Time:2019-03-11  Hits:

Indexed by: Journal Article

Date of Publication: 2015-01-01

Journal: ICIC Express Letters

Included Journals: Scopus、EI

Volume: 9

Issue: 4

Page Number: 1143-1151

ISSN: 1881803X

Abstract: Collaborative filtering algorithm is the most widely used algorithm in personalized recommendation field. However, with the advent of big data era, it will be very time-consuming when the recommendation system becomes large-scale. Currently, the main method to improve the efficiency is using distributed cluster, but it may not be accessible to most researchers due to its high cost. This paper proposes a multi-GPU (Graphics Processing Unit) based parallel collaborative filtering algorithm to improve the efficiency and scalability. Experiment result shows that multi-GPU can significantly improve the running speed and the larger scale the problem is, the higher the speedup is, which provides a feasible solution to save cost for SMEs (Small and Medium-sized Enterprises). ?, 2015 ICIC international.