Learning Similarity Measures from Pairwise Constraints with Neural Networks

Authors:
Marco Maggini;Stefano Melacci;Lorenzo Sarti
Affiliations:
DII, Università degli Studi di Siena, Siena, (Italy) 53100;DII, Università degli Studi di Siena, Siena, (Italy) 53100;DII, Università degli Studi di Siena, Siena, (Italy) 53100
Venue:
ICANN '08 Proceedings of the 18th international conference on Artificial Neural Networks, Part II
Year:
2008

Citing 7
Cited 1

Original Contribution: Multilayer feedforward networks with a nonpolynomial activation function can approximate any function

Neural Networks
Similarity Measures

IEEE Transactions on Pattern Analysis and Machine Intelligence
A probabilistic framework for semi-supervised clustering

Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining
Integrating constraints and metric learning in semi-supervised clustering

ICML '04 Proceedings of the twenty-first international conference on Machine learning
Boosting margin based distance functions for clustering

ICML '04 Proceedings of the twenty-first international conference on Machine learning
Learning a Mahalanobis Metric from Equivalence Constraints

The Journal of Machine Learning Research
Learning a kernel function for classification with small training samples

ICML '06 Proceedings of the 23rd international conference on Machine learning

Learning from pairwise constraints by Similarity Neural Networks

Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a novel neural network model, called Similarity Neural Network (SNN), designed to learn similarity measures for pairs of patterns exploiting binary supervision. The model guarantees to compute a non negative and symmetric measure, and shows good generalization capabilities even if a small set of supervised examples is used for training. The approximation capabilities of the proposed model are also investigated. Moreover, the experiments carried out on some benchmark datasets show that SNNs almost always outperform other similarity learning methods proposed in the literature.