Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques

Details Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques

Dec 2006

adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2006agufmsm51a1392d&link_type=abstract

American Geophysical Union, Fall Meeting 2006, abstract #SM51A-1392

Mathematics

Logic

0520 Data Analysis: Algorithms And Implementation, 0555 Neural Networks, Fuzzy Logic, Machine Learning, 7526 Magnetic Reconnection (2723, 7835), 7835 Magnetic Reconnection (2723, 7526)

Scientific paper

We report preliminary results concerning the combined use of unsupervised and supervised techniques to classify Geotail FTEs. Currently, humans identify FTEs on the basis of clear isolated bipolar signatures normal to the nominal magnetopause, magnetic field strength enhancements, and sometimes east/west deflections of the magnetic field in the plane of the magnetopause BM. However, events with decreases or crater-like structures in the magnetic field strength, no east/west deflection, and asymmetric or continuous variations normal to the magnetopause have also been identified as FTEs, making statistical studies of FTEs problematical. Data mining techniques are particularly useful in developing automated search algorithms and generating large event lists for statistical studies. Data mining techniques can be divided into two types, supervised and unsupervised. In supervised algorithms, one teaches the algorithm using examples from labeled data. Considering the case of FTEs, the user would provide examples of FTEs as well as examples of non-FTEs and label (as FTE or non-FTE) the data. Since one has to start with a labeled data set, this may already include a user bias in the selection process. To avoid this issue, it can be useful to employ unsupervised techniques. Unsupervised techniques are analogous to training without a teacher: data are not labeled. There is also hybrid modeling where one makes several models, using unsupervised and supervised techniques and then connects them into a hybrid model.

Affiliated with

Also associated with

No associations

Rating

Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques 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 Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Statistical Studies of Flux Transfer Events Using Unsupervised and Supervised Techniques will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-975018