Towards Optimal Model Order Selection for Autoregressive Spectral Analysis of Mental Tasks Using Genetic Algorithm

Summary Autoregressive (AR) models for spectral analysis of electroencephalogram (EEG) signals are advantageous over the classical Fourier transform methods due to their ability to deal with short segments of data, superior resolution and smoother spectra. But a problem or rather a parameter that needs to be optimised in this method is the model order of the AR equation. Currently, statistical methods like Akaike Information Criterion, Final Prediction Error, Residual Variance, Minimum Description Length, Criterion Autoregressive Transfer and Hannan-Quinn have been used for this purpose. These methods depend on the statistical properties of the data, which selects the lowest order that is optimal to represent the signal. In this paper, the use of genetic algorithm (GA) is proposed to select the order of the AR model during classifier training. This technique fuses GA with Fuzzy ARTMAP to select the appropriate AR model order for EEG signals during system training to optimise classification of test signals into their respective different mental tasks. The experimental results show that this method outperforms the other statistical methods and a fixed 6th order model although the simulations were carried out with a small number of genetic populations and generations to reduce the computational cost.

[1]  T. Ulrich,et al.  Maximum entropy spectral analy-sis and autoregressive decomposition , 1975 .

[2]  J. Bronzino,et al.  Autoregressive and bispectral analysis techniques: EEG applications , 1990, IEEE Engineering in Medicine and Biology Magazine.

[3]  David E. Goldberg,et al.  Genetic Algorithms in Search Optimization and Machine Learning , 1988 .

[4]  George E. P. Box,et al.  Time Series Analysis: Forecasting and Control , 1977 .

[5]  D. G. Watts,et al.  Spectral analysis and its applications , 1968 .

[6]  Donald G. Childers,et al.  Modern Spectrum Analysis , 1978 .

[7]  E. Parzen Some recent advances in time series modeling , 1974 .

[8]  R. F. Wagner,et al.  A comparison of autoregressive spectral estimation algorithms and order determination methods in ultrasonic tissue characterization , 1995, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control.

[9]  Brahim Aksasse,et al.  Two-dimensional autoregressive (2-D AR) model order estimation , 1999, IEEE Trans. Signal Process..

[10]  Z. Keirn,et al.  A new mode of communication between man and his surroundings , 1990, IEEE Transactions on Biomedical Engineering.

[11]  Stephen Grossberg,et al.  A fuzzy ARTMAP nonparametric probability estimator for nonstationary pattern recognition problems , 1995, IEEE Trans. Neural Networks.

[13]  J. Rissanen,et al.  Modeling By Shortest Data Description* , 1978, Autom..

[14]  H. Akaike A new look at the statistical model identification , 1974 .

[15]  Richard H. Jones Autoregression order selection , 1976 .

[16]  Stephen Grossberg,et al.  Fuzzy ARTMAP: A neural network architecture for incremental supervised learning of analog multidimensional maps , 1992, IEEE Trans. Neural Networks.

[17]  Seiji Inokuchi,et al.  An Approach to Analysis of EEGs Recorded During Music Listening , 1997 .

[18]  A.F.C. Infantosi,et al.  On the selection of autoregressive order for electroencephalographic (EEG) signals , 1995, 38th Midwest Symposium on Circuits and Systems. Proceedings.

[19]  Richard H. Jones Identification and autoregressive spectrum estimation , 1974 .

[20]  Richard Aufrichtig,et al.  Order estimation and model verification in autoregressive modeling of EEG sleep recordings , 1992, 1992 14th Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[21]  D. E. Goldberg,et al.  Genetic Algorithm in Search , 1989 .

[22]  J. Millis,et al.  THE UNIVERSITY OF , 2000 .

[23]  R. Lacoss,et al.  Some Geophysical Applications of Autoregressive Spectral Estimates , 1977, IEEE Transactions on Geoscience Electronics.

[24]  G Pfurtscheller,et al.  Separability of EEG signals recorded during right and left motor imagery using adaptive autoregressive parameters. , 1998, IEEE transactions on rehabilitation engineering : a publication of the IEEE Engineering in Medicine and Biology Society.

[25]  E. A. Stolz,et al.  Multivariate Autoregressive Models for Classification of Spontaneous Electroencephalogram During Mental Tasks1 , 1998 .

[26]  Ben H. Jansen,et al.  Autoregressive Estimation of Short Segment Spectra for Computerized EEG Analysis , 1981, IEEE Transactions on Biomedical Engineering.

[27]  P. Young,et al.  Time series analysis, forecasting and control , 1972, IEEE Transactions on Automatic Control.

[28]  C.W. Anderson,et al.  Multivariate autoregressive models for classification of spontaneous electroencephalographic signals during mental tasks , 1998, IEEE Transactions on Biomedical Engineering.

[29]  H. Jasper,et al.  The ten-twenty electrode system of the International Federation. The International Federation of Clinical Neurophysiology. , 1999, Electroencephalography and clinical neurophysiology. Supplement.

[30]  Stephen Grossberg,et al.  A massively parallel architecture for a self-organizing neural pattern recognition machine , 1988, Comput. Vis. Graph. Image Process..

[31]  John H. Holland,et al.  Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence , 1992 .

[32]  Mingui Sun,et al.  Vector autoregressive model selection in multichannel EEG , 1997, Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136).

[33]  On the use of the Akaike Information Criterion in AR spectral analysis of cardiovascular variability signals: a case report study , 1993, Proceedings of Computers in Cardiology Conference.

[34]  B. G. Quinn,et al.  The determination of the order of an autoregression , 1979 .