A GPU-based parallel algorithm for time series pattern mining

Tao Sun, Jian Sha, Lin Feng
School of Innovation Experiment, DaLian University of Technology, Dalian, China
Journal of Convergence Information Technology (JCIT), Vol. 6, No. 12, pp. 163 – 170, 2011


   title={A GPU-based parallel algorithm for time series pattern mining},

   author={Sun, Tao and Sha, Jian and Feng, Lin},



Download Download (PDF)   View View   Source Source   



Mining of time series pattern is an important research area, of which getting LCSS(Longest Common Subsequence) between high-dimensional time series is one of the most important issues. Large scale data needs to be handled in practical applications, so the research of efficient retrieval method is becoming a realistic work. Based on the issues above, we propose an efficient parallel algorithm to get LCSS between time series with the help of GPU (Graphics Processor Unit). On that basis, propose a parallel limit least matching rate LCSS algorithm (Parallel-Limited-LCSS), and optimize the retrieve parts of the algorithm with the help of inverted index structure, so as to enhance the efficiency of the algorithm. Experiments show that our algorithm has excellent speed and accuracy, and can be applied to the field of data mining widely.
No votes yet.
Please wait...

* * *

* * *

HGPU group © 2010-2021 hgpu.org

All rights belong to the respective authors

Contact us: