Computer Science – Learning
Scientific paper
Dec 2001
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2001agufmsm31a0748g&link_type=abstract
American Geophysical Union, Fall Meeting 2001, abstract #SM31A-0748
Computer Science
Learning
2788 Storms And Substorms, 7839 Nonlinear Phenomena, 7894 Instruments And Techniques
Scientific paper
Reliable forecasting of rare events in a complex dynamical system is a challenging problem that is important for many practical applications. Due to the nature of rare events, data set available for construction of the statistical and/or machine learning model is often very limited and incomplete. Therefore many widely used approaches including such robust algorithms as neural networks can easily become inadequate for rare events prediction. Moreover in many practical cases models with high-dimensional inputs are required. This limits applications of the existing rare event modeling techniques (e.g., extreme value theory) that focus on univariate cases. These approaches are not easily extended to multivariate cases. Support vector machine (SVM) is a machine learning system that can provide an optimal generalization using very limited and incomplete training data sets and can efficiently handle high-dimensional data. These features may allow to use SVM to model rare events in some applications. We have applied SVM-based system to the problem of large-amplitude substorm prediction and extreme event forecasting in stock and currency exchange markets. Encouraging preliminary results will be presented and other possible applications of the system will be discussed.
Ganguli Supriya B.
Gavrishchaka Valeriy V.
No associations
LandOfFree
Rare events modeling with support vector machine: Application to forecasting large-amplitude geomagnetic substorms and extreme events in financial markets. does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Rare events modeling with support vector machine: Application to forecasting large-amplitude geomagnetic substorms and extreme events in financial markets., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rare events modeling with support vector machine: Application to forecasting large-amplitude geomagnetic substorms and extreme events in financial markets. will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1384049