Kernel matching pursuit for large datasets

POPOVICI, Vlad, S BENGIO a JP THIRAN. Kernel matching pursuit for large datasets. PATTERN RECOGNITION. OXFORD: PERGAMON-ELSEVIER SCIENCE LTD, 2005, roč. 38, č. 12, s. 2385-2390. ISSN 0031-3203. Dostupné z: https://dx.doi.org/10.1016/j.patcog.2005.01.021.

Základní údaje

• Originální název: Kernel matching pursuit for large datasets
• Autoři: POPOVICI, Vlad, S BENGIO a JP THIRAN.
• Vydání: PATTERN RECOGNITION, OXFORD, PERGAMON-ELSEVIER SCIENCE LTD, 2005, 0031-3203.

Další údaje

• Originální jazyk: angličtina
• Typ výsledku: Článek v odborném periodiku
• Utajení: není předmětem státního či obchodního tajemství
• Impakt faktor: Impact factor: 2.153
• Doi: http://dx.doi.org/10.1016/j.patcog.2005.01.021
• UT WoS: 000232703000014
• Klíčová slova anglicky: kernel matching pursuit; greedy algorithm; sparse classifier

Anotace

Kernel matching pursuit is a greedy algorithm for building an approximation of a discriminant function as a linear combination of some basis functions selected from a kernel-induced dictionary. Here we propose a modification of the kernel matching pursuit algorithm that aims at making the method practical for large datasets. Starting from an approximating algorithm, the weak greedy algorithm, we introduce a stochastic method for reducing the search space at each iteration. Then we study the implications of using an approximate algorithm and we show how one can control the trade-off between the accuracy and the need for resources. Finally, we present some experiments performed on a large dataset that support our approach and illustrate its applicability. (c) 2005 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.