An Online Metric Learning Approach through Margin Maximization

Pérez-Suay, Adrián; Ferri, Francesc J.; Albert, Jesús V.

doi:10.1007/978-3-642-21257-4_62

Cited by 4 publications

(6 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alternative formulations of this idea have already been proposed and studied [10], [11], [15]. As the purpose of the present work is to compare batch and online formulations, only the case of quadratic penalties has been considered.…”

Section: B Online Formulationmentioning

confidence: 99%

“…Instead of considering constrained optimization using all available instances, the DML problem can be solved in a more convenient way both from the point of view of computation and robustness by using an online learning approach [10], [15]. At each time step, k, a new optimization problem is formulated and solved using a single particular instance (a labeled pair, (x i , x j ), and their corresponding labels, (c i , c j )), that is made available to the system.…”

Section: B Online Formulationmentioning

confidence: 99%

See 1 more Smart Citation

Comparative Evaluation of Batch and Online Distance Metric Learning Approaches Based on Margin Maximization

Pérez-Suay

Ferri

Arevalillo‐Herráez

et al. 2013

2013 IEEE International Conference on Systems, Man, and Cybernetics

Self Cite

View full text Add to dashboard Cite

Distance metric learning aims at obtaining an appropriate metric that conveniently adapts to a particular recognition problem given a set of training pairs. The idea of maximizing a margin that separates similar and dissimilar objects has been used in different ways in several recent works. This paper considers two different learning schemes aiming at the same goal but posing the learning problem either as a batch or as an online formulation. Extensive experiments and the corresponding discussion try to put forward the advantages and drawbacks of each of the approaches considered.

show abstract

Section: B Online Formulationmentioning

confidence: 99%

Section: B Online Formulationmentioning

confidence: 99%

Comparative Evaluation of Batch and Online Distance Metric Learning Approaches Based on Margin Maximization

Pérez-Suay

Ferri

Arevalillo‐Herráez

et al. 2013

2013 IEEE International Conference on Systems, Man, and Cybernetics

Self Cite

View full text Add to dashboard Cite

show abstract

Section: Online Formulation Using Marginsmentioning

confidence: 99%

“…The set is given to the algorithm at least twice and then this process is repeated until a maximum number of iterations has been reached. This maximum number of iterations has been established as the minimum between 20 % of all possible pairs of training samples (as in previous studies [17]) and 50 times the number of pairs in P (which is by far more than enough for larger databases with small number of classes). This means that at least t steps are executed, with t = min(…”

Section: Experimental Settingsmentioning

confidence: 99%

Passive–aggressive online distance metric learning and extensions

2013

Self Cite

View full text Add to dashboard Cite

Online metric learning aims at constructing an appropriate dissimilarity measure from training data composed of labeled pairs of samples. Margin maximization has been widely used as an efficient approach to address this problem. This paper reviews various existing online metric learning formulations, and also introduces an alternative passive-aggressive scheme. In addition, the pros and cons of each alternative are analyzed in a comparative empirical study over several databases.

show abstract

“…Instead of considering constrained optimization using all information (examples) available, the above problem can be solved in a more convenient way both from the point of view of computation and robustness by using an online learning approach [4,8] . Under this sequential scheme, at each step k, a particular model formed by the pair (M k , b k ) is available to make a prediction over the labeled pair t k = (x i , x j , y ij ) which is revealed to the system at this step.…”

Section: The Separable Casementioning

confidence: 99%

Online Metric Learning Methods Using Soft Margins and Least Squares Formulations

Pérez-Suay

Ferri

2012

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. Online metric learning using margin maximization has been introduced as a way to learn appropriate dissimilarity measures in an efficient way when information as pairs of examples is given to the learning system in a progressive way. These schemes have several practical advantages with regard to global ones in which a training set needs to be processed. On the other hand, they may suffer from a poor performance depending on the quality of the examples and the particular tuning or other implementation details. This paper formulates several online metric learning alternatives using a passive-aggressive schema. A new formulation of the online problem using least squares is also introduced. The relative behavior of the different alternatives is studied and comparative experimentation is carried out to put forward the benefits and weaknesses of each alternative.

show abstract

An Online Metric Learning Approach through Margin Maximization

Cited by 4 publications

References 4 publications

Comparative Evaluation of Batch and Online Distance Metric Learning Approaches Based on Margin Maximization

Comparative Evaluation of Batch and Online Distance Metric Learning Approaches Based on Margin Maximization

Passive–aggressive online distance metric learning and extensions

Online Metric Learning Methods Using Soft Margins and Least Squares Formulations

Contact Info

Product

Resources

About