Semi-supervised Kernel Metric Learning Using Relative Comparisons

By Ehsan Amid, Aristides Gionis, Antti Ukkonen. 2016

10 citations recorded by the source corpus.

We consider the problem of metric learning subject to a set of constraints on relative-distance comparisons between the data items. Such constraints are meant to reflect side-information that is not expressed directly in the feature vectors of the data items. The relative-distance constraints used in this work are particularly effective in expressing structures at finer level of detail than must-link (ML) and cannot-link (CL) constraints, which are most commonly used for semi-supervised clustering. Relative-distance constraints are thus useful in settings where providing an ML or a CL constraint is difficult because the granularity of the true clustering is unknown. Our main contribution is an efficient algorithm for learning a kernel matrix using the log determinant divergence --- a variant of the Bregman divergence --- subject to a set of relative-distance constraints. The learned kernel matrix can then be employed by many different kernel methods in a wide range of applications. In our experimental evaluations, we consider a semi-supervised clustering setting and show empirically that kernels found by our algorithm yield clusterings of higher quality than existing approaches that either use ML/CL constraints or a different means to implement the supervision using relative comparisons.

How researchers see this paper

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

Citations in the corpus

TriMap: Large-scale Dimensionality Reduction Using Triplets (2019, methodology): “The key idea behind TriMap stems from semi-supervised metric learning (Amid et al., 2016): Given an initial low-dimensional representation for the data points, the triplet information from the highdimensional representation of the points is used to enhance the quality of the embedding.”
Learning to cluster in order to Transfer across domains and tasks (2017, methodology): “, 2014), SKKm (Anand et al., 2014; Amid et al., 2016), and SKLR (Amid et al.”
Multi-class Classification without Multi-class Labels (2019, unknown): “…(Chen & Cai, 2011) 23.6% 18.0% 0.376 0.500 ITML (Davis et al., 2007) 56.7% 47.2% 0.674 0.727 SKKm (Anand et al., 2014) 62.4% 46.9% 0.770 0.781 SKLR (Amid et al., 2016) 66.9% 46.8% 0.791 0.760 CSP (Wang et al., 2014) 62.5% 65.4% 0.812 0.812 MPCK-means (Bilenko et al., 2004) We use constrained…”