Investigating the impact of hubness on SVM classifiers : μεταπτυχιακή διατριβή
Support Vector Machines is a well known classification algorithm. Its very good performance has been proven theoretically and observed in real life data sets. Recent works have demonstrated the existence of hubs in high dimensional spaces. Hubs are points that are very popular among other points. Th...
Αποθηκεύτηκε σε:
| Κύριος συγγραφέας: | |
|---|---|
| Συγγραφή απο Οργανισμό/Αρχή: | |
| Μορφή: | Thesis Βιβλίο |
| Γλώσσα: | English |
| Δημοσίευση: |
2011.
|
| Θέματα: | |
| Διαθέσιμο Online: | http://hdl.handle.net/11610/12556 |
| Ετικέτες: |
Προσθήκη ετικέτας
Δεν υπάρχουν, Καταχωρήστε ετικέτα πρώτοι!
|
| Περίληψη: | Support Vector Machines is a well known classification algorithm. Its very good performance has been proven theoretically and observed in real life data sets. Recent works have demonstrated the existence of hubs in high dimensional spaces. Hubs are points that are very popular among other points. That is in a set of data points in a high dimensional space, a big number of these have the hub points as their nearest neighbors. This work concentrates on the study of the relation between hubs and support vector machines and especially on the differences between high and low dimensionality. |
|---|---|
| Περιγραφή τεκμηρίου: | Μέλη της εξεταστικής επιτροπής: Ευστάθιος Σταματάτος, Αλέξανδρος Νανόπουλος, Γεώργιος Βούρος. |
| Φυσική περιγραφή: | 59 σ. : εικ., πιν. ; 3ο εκ. |
| Βιβλιογραφία: | Βιβλιογραφία: σ. 58-59. |
| Πρόσβαση: | Διάθεση πλήρους κειμένου - Ελεύθερη πρόσβαση. |