Learning Distance Metrics with Contextual Constraints for Image Retrieval

By Steven C. H. Hoi, Wei David Liu, Michael R. Lyu, Wei-Ying Ma. Computer Vision and Pattern Recognition, 2006

326 citations recorded by the source corpus.

Relevant Component Analysis (RCA) has been proposed for learning distance metrics with contextual constraints for image retrieval. However, RCA has two important disadvantages. One is the lack of exploiting negative constraints which can also be informative, and the other is its incapability of capturing complex nonlinear relationships between data instances with the contextual information. In this paper, we propose two algorithms to overcome these two disadvantages, i.e., Discriminative Component Analysis (DCA) and Kernel DCA. Compared with other complicated methods for distance metric learning, our algorithms are rather simple to understand and very easy to solve. We evaluate the performance of our algorithms on image retrieval in which experimental results show that our algorithms are effective and promising in learning good quality distance metrics for image retrieval.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/77f0d765e9557c8260bb39087985103527f7c702

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Learning for Content-Based Image Retrieval: A Comprehensive Study (2014, background): “In terms of different learning approaches, distance metric learning techniques are typically categorized into two groups: the global supervised approaches [2, 18] that learn a metric on a global setting by satisfying all the constraints simultaneously, the local supervised approaches [48, 12] that l”
Cross-Modal Similarity Learning: A Low Rank Bilinear Formulation (2014, methodology): “ysis (RCA) [2], and the Discriminative Component Analysis (DCA) which is based on the RCA [16].”
Online multimodal deep similarity learning with application to image retrieval (2013, methodology): “…in terms of training data formats, many DML studies in machine learning typically learn metrics directly from explicit class labels [47], while DML studies in multimedia mainly learn metrics from side information in the forms of either pairwise constraints [16, 15] or triplet constraints [8].”
Regularized Distance Metric Learning: Theory and Algorithm (2009, background): “It is critical to many real-world applications, such as information retrieval, classification, and clustering [6, 7].”
Metric Learning: A Survey (2013, background): “Finally, we note that [32] provides another example of applying metric learning to the problem of image retrieval.”
Learning a Mahalanobis distance metric for data clustering and classification (2008, background): “The results obtained by standard Kmeans, Kmeans+Xing's method, Kmeans+RCA, Kmeans+DCA and Kmeans+our-method are reported in Table 4.”
On Deep Multi-View Representation Learning (2015, background): “…which may be known a priori or derived from class labels (Xing et al., 2003; Shental et al., 2002; Bar-Hillel et al., 2005; Tsang and Kwok, 2003; Schultz and Joachims, 2004; Goldberger et al., 2005; Hoi et al., 2006; Globerson and Roweis, 2006; Davis et al., 2007; Weinberger and Saul, 2009).”
Deep Multimodal Distance Metric Learning Using Click Constraints for Image Ranking (2017, methodology): “The discriminative component analysis (DCA) is proposed in [49], and they use two groups of pairwise constraints of the labeled data to train the distance metric.”
Natural Language Object Retrieval (2015, methodology): “In image retrieval, a ranking function is learned through a recurrent neural network [24, 4], metric learning [13], correlation analysis [20] and other methods [6, 19].”
Large scale metric learning from equivalence constraints (2012, background): “It can significantly improve results for tracking [22], image retrieval [11], face identification [9], clustering [21], or person re-identification [4].”