We propose a manifold regularization algorithm designed to work in an on-line scenario where data arrive continuously over time and it is not feasible to completely store the data stream for training the classifier in batch mode. The On-line Laplacian One-Class SVM (OLapOCSVM) algorithm exploits both positively labeled and totally unlabeled examples, updating the classifier hypothesis as new data becomes available. The learning procedure is based on conjugate gradient descent in the primal formulation of the SVM. The on-line algorithm uses an efficient buffering technique to deal with the continuous incoming data. In particular, we define a buffering policy that is based on the current estimate of the support of the input data distribution. The experimental results on real-world data show that OLapOCSVM compares favorably with the corresponding batch algorithms, while making it possible to be applied in generic on-line scenarios with limited memory requirements. © 2013 Springer-Verlag Berlin Heidelberg.

On-line laplacian one-class support vector machines / LIPPI, MARCO. - ELETTRONICO. - 8131:(2013), pp. 186-193. (Intervento presentato al convegno 23rd International Conference on Artificial Neural Networks, ICANN 2013 tenutosi a Sofia, bgr nel 2013) [10.1007/978-3-642-40728-4_24].

On-line laplacian one-class support vector machines

LIPPI, MARCO
2013

Abstract

We propose a manifold regularization algorithm designed to work in an on-line scenario where data arrive continuously over time and it is not feasible to completely store the data stream for training the classifier in batch mode. The On-line Laplacian One-Class SVM (OLapOCSVM) algorithm exploits both positively labeled and totally unlabeled examples, updating the classifier hypothesis as new data becomes available. The learning procedure is based on conjugate gradient descent in the primal formulation of the SVM. The on-line algorithm uses an efficient buffering technique to deal with the continuous incoming data. In particular, we define a buffering policy that is based on the current estimate of the support of the input data distribution. The experimental results on real-world data show that OLapOCSVM compares favorably with the corresponding batch algorithms, while making it possible to be applied in generic on-line scenarios with limited memory requirements. © 2013 Springer-Verlag Berlin Heidelberg.
2013
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
23rd International Conference on Artificial Neural Networks, ICANN 2013
Sofia, bgr
2013
LIPPI, MARCO
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificatore per citare o creare un link a questa risorsa: https://hdl.handle.net/2158/1356477
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? ND
social impact