An efficient sparse metric learning in high-dimensional space via l1-penalized log-determinant regularization

Authors:
Guo-Jun Qi;Jinhui Tang;Zheng-Jun Zha;Tat-Seng Chua;Hong-Jiang Zhang
Affiliations:
University of Illinois at Urbana-Champaign, Urbana, IL;National University of Singapore, Singapore;National University of Singapore, Singapore;National University of Singapore, Singapore;Microsoft Advanced Technology Center, Beijing, China
Venue:
ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
Year:
2009

Citing 4
Cited 4

Information-theoretic metric learning

Proceedings of the 24th international conference on Machine learning
Optimal dimensionality of metric space for classification

Proceedings of the 24th international conference on Machine learning
Model Selection Through Sparse Maximum Likelihood Estimation for Multivariate Gaussian or Binary Data

The Journal of Machine Learning Research
Robust Face Recognition via Sparse Representation

IEEE Transactions on Pattern Analysis and Machine Intelligence

Semi-supervised sparse metric learning using alternating linearization optimization

Proceedings of the 16th ACM SIGKDD international conference on Knowledge discovery and data mining
Learning similarity function for rare queries

Proceedings of the fourth ACM international conference on Web search and data mining
Fast neighborhood component analysis

Neurocomputing
Clustering tagged documents with labeled and unlabeled documents

Information Processing and Management: an International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes an efficient sparse metric learning algorithm in high dimensional space via an l1-penalized log-determinant regularization. Compare to the most existing distance metric learning algorithms, the proposed algorithm exploits the sparsity nature underlying the intrinsic high dimensional feature space. This sparsity prior of learning distance metric serves to regularize the complexity of the distance model especially in the "less example number p and high dimension d" setting. Theoretically, by analogy to the covariance estimation problem, we find the proposed distance learning algorithm has a consistent result at rate O (√m2 log d)/n) to the target distance matrix with at most m nonzeros per row. Moreover, from the implementation perspective, this l1-penalized log-determinant formulation can be efficiently optimized in a block coordinate descent fashion which is much faster than the standard semi-definite programming which has been widely adopted in many other advanced distance learning algorithms. We compare this algorithm with other state-of-the-art ones on various datasets and competitive results are obtained.