Support vector machine as an efficient tool for high‐dimensional data processing: Application to substorm forecasting

The support vector machine (SVM) has been used to model solar wind-driven geomagnetic substorm activity characterized by the auroral electrojet (AE) index. The focus of the present study, which is the first application of the SVM to space physics problems, is reliable prediction of large-amplitude substorm events from solar wind and interplanetary magnetic field data. This forecasting problem is important for many practical applications as well as for further understanding of the overall substorm dynamics. SVM has been trained on symbolically encoded AE index time series to perform supercritical/subcritical classification with respect to an application-dependent threshold. It is shown that SVM performance can be comparable to or even superior to that of the neural networks model. The advantages of the SVM-based techniques are expected to be much more pronounced in future space weather forecasting models, which will incorporate many types of high-dimensional, multiscale input data once real time availability of this information becomes technologically feasible.

[1]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[2]  Vladimir Vapnik,et al.  Chervonenkis: On the uniform convergence of relative frequencies of events to their probabilities , 1971 .

[3]  F ROSENBLATT,et al.  The perceptron: a probabilistic model for information storage and organization in the brain. , 1958, Psychological review.

[4]  A. Sharma,et al.  Assessing the magnetosphere's nonlinear behavior: Its dimension is low, its predictability, high , 1995 .

[5]  Wendell Horton,et al.  A Low-Dimensional Dynamical Model for the Solar Wind Driven Geotail-Ionosphere System , 1998 .

[6]  H. Gleisner,et al.  Response of the auroral electrojets to the solar wind modeled with neural networks , 1997 .

[7]  C. Mobarry,et al.  Topological structure of the magnetotail as a function of interplanetary magnetic field direction , 1995 .

[8]  Toshiki Tajima,et al.  Forecasting auroral electrojet activity from solar wind input with neural networks , 1999 .

[9]  Christopher M. Bishop,et al.  Neural networks for pattern recognition , 1995 .

[10]  Vladimir N. Vapnik,et al.  The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[11]  R Urbanczik,et al.  Universal learning curves of support vector machines. , 2001, Physical review letters.

[12]  Jose C. Principe,et al.  Neural and adaptive systems , 2000 .

[13]  J. P. Smith,et al.  The solar-wind driven magnetosphere–ionosphere as a complex dynamical system , 1999 .

[14]  Tom Chang,et al.  Self-organized criticality, multi-fractal spectra, sporadic localized reconnections and intermittent turbulence in the magnetotail , 1999 .

[15]  Nello Cristianini,et al.  An Introduction to Support Vector Machines and Other Kernel-based Learning Methods , 2000 .

[16]  J. Joselyn.  Geomagnetic activity forecasting: The state of the art , 1995 .

[17]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .

[18]  Daniel N. Baker,et al.  The organized nonlinear dynamics of the magnetosphere , 1996 .

[19]  R. Schunk,et al.  Global ionosphere‐polar wind system during changing magnetic activity , 1997 .

[20]  Valeriy V. Gavrishchaka,et al.  Optimization of the neural-network geomagnetic model for forecasting large-amplitude substorm events , 2001 .

[21]  Chih-Jen Lin,et al.  The analysis of decomposition methods for support vector machines , 2000, IEEE Trans. Neural Networks Learn. Syst..

[22]  H. Gleisner,et al.  Predicting geomagnetic storms from solar-wind data using time-delay neural networks , 1996 .