Title :
Modeling the Dynamics of the Human Pulse Data by MDL-optimal Neural Networks
Author :
Ma, Yingnan ; Zhao, Yi ; Fan, Youhua ; Hu, Hong ; Zhang, Xiujun
Author_Institution :
Shenzhen Grad. Sch., Harbin Inst. of Technol., Shenzhen
Abstract :
In this paper, we describe an information theoretic criterion, the method of minimum description length (MDL), to determine optimal neural networks to predict the human pulse data as well as non-stationary Lorenz data. Such optimal models which minimize the description length of the data both generalize well and accurately capture the dynamics of the original data. It demonstrates the potential utility of our MDL-optimal model in biomedical time series modeling.
Keywords :
information theory; medical signal processing; neural nets; time series; biomedical time series; human pulse dynamics; information theoretic criterion; minimum description length; nonstationary Lorenz data; optimal neural networks; Artificial neural networks; Biomedical engineering; Biomedical informatics; Costs; Humans; Information science; Neural networks; Neurons; Predictive models; Testing; Human Pulse Data; MDL-optimal;
Conference_Titel :
BioMedical Engineering and Informatics, 2008. BMEI 2008. International Conference on
Conference_Location :
Sanya
Print_ISBN :
978-0-7695-3118-2
DOI :
10.1109/BMEI.2008.74
