doi: 10.18178/ijfcc.2017.6.3.503
A Deep Learning of Time Series for Efficient Analysis
Tippaya Thinsungnoen
*
,
Kittisak Kerdprasop
,
Nittaya Kerdprasop
- School of Computer Engineering, Suranaree University of Technology (SUT), 111 University Avenue, Muang, Nakhon Ratchasima 30000, Thailand.
Abstract
The main problem for analyzing time series datawith machine learning techniques such as classification andclustering is that a high-dimensional nature of this kind of datacan cause computational difficulty in finding optimal solution.Currently, advanced learning strategy such as deep learninghas been used extensively and effectively to improve learningperformance. In this research, we propose a method to optimizetime series analysis by adding a pre-training and fine-tuningprocess of deep learning based on Deep Belief Networks andRestricted Boltzmann Machines. On evaluating performance ofthe proposed method, we use electroencephalographic,electrocardiogram, and synthetic time series data to analyzewith classification task. The induced classification models areassessed with the four several metrics including clusterevaluation, purity, mean squared error, and processing time.We comparatively compare the three learning schemes:traditional neural networks, deep learning networks, and deeplearning networks with added a pre-training and fine-tuningprocess. The results showed that all three schemes show thesame performance on predicting time series data when assessedwith mean squared error. For the processing time comparison,neural networks technique is slightly faster than others. Butwhen assessed with cluster formation and purity metrics, wefound that deep learning based on the concept of Deep BeliefNetworks and Restricted Boltzmann Machines that adds apre-training and fine-tuning process outperforms otherlearning techniques.
Keywords
- Deep belief networks
- deep learning
- restrictedBoltzmann machines
- time series analysis
How to Cite
Tippaya Thinsungnoen, Kittisak Kerdprasop, and Nittaya Kerdprasop, "A Deep Learning of Time Series for Efficient Analysis," International Journal of Future Computer and Communication, vol. 6, no. 3, pp. 123-127, 2017. https://doi.org/ 10.18178/ijfcc.2017.6.3.503
Copyright & License
Copyright © 2017 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).