Online learning algorithms are fast, memory-efficient, easy to implement, and applicable to many prediction problems, including classification, regression, and ranking. Several online algorithms were proposed in the past few decades, some based on additive updates, like the Perceptron, and some on multiplicative updates, like Winnow. A unifying perspective on the design and the analysis of online algorithms is provided by online mirror descent, a general prediction strategy from which most first-order algorithms can be obtained as special cases. We generalize online mirror descent to time-varying regularizers with generic updates. Unlike standard mirror descent, our more general formulation also captures second order algorithms, algorithms for composite losses and algorithms for adaptive filtering. Moreover, we recover, and sometimes improve, known regret bounds as special cases of our analysis using specific regularizers. Finally, we show the power of our approach by deriving a new second order algorithm with a regret bound invariant with respect to arbitrary rescalings of individual features.

A generalized online mirror descent with applications to classification and regression / F. Orabona, K. Crammer, N. Cesa-Bianchi. - In: MACHINE LEARNING. - ISSN 0885-6125. - 99:3(2015 Jun), pp. 411-435.

A generalized online mirror descent with applications to classification and regression

F. Orabona
;
N. Cesa-Bianchi
Ultimo
2015

Abstract

Online learning algorithms are fast, memory-efficient, easy to implement, and applicable to many prediction problems, including classification, regression, and ranking. Several online algorithms were proposed in the past few decades, some based on additive updates, like the Perceptron, and some on multiplicative updates, like Winnow. A unifying perspective on the design and the analysis of online algorithms is provided by online mirror descent, a general prediction strategy from which most first-order algorithms can be obtained as special cases. We generalize online mirror descent to time-varying regularizers with generic updates. Unlike standard mirror descent, our more general formulation also captures second order algorithms, algorithms for composite losses and algorithms for adaptive filtering. Moreover, we recover, and sometimes improve, known regret bounds as special cases of our analysis using specific regularizers. Finally, we show the power of our approach by deriving a new second order algorithm with a regret bound invariant with respect to arbitrary rescalings of individual features.
relative loss bounds; learning algorithms; convex-optimization; subgradient methods
Settore INF/01 - Informatica
giu-2015
25-dic-2014
Article (author)
File in questo prodotto:
File Dimensione Formato  
1304.2994.pdf

accesso riservato

Tipologia: Pre-print (manoscritto inviato all'editore)
Dimensione 633.8 kB
Formato Adobe PDF
633.8 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
art%3A10.1007%2Fs10994-014-5474-8.pdf

accesso riservato

Tipologia: Publisher's version/PDF
Dimensione 783.82 kB
Formato Adobe PDF
783.82 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
Pubblicazioni consigliate

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

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/285628
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 45
  • ???jsp.display-item.citation.isi??? 31
social impact