Coordinated Local Metric Learning Shreyas Saxena, Jakob Verbeek

As mentioned in 4.2, our algorithm solves an inverse problem: given a sequence of histograms representing a movement of mass, we aim at fitting a metric for which that sequence can

In this paper, we design a new local Mahalanobis-like metric learning algorithm that aims at approximating a perceptual scene color difference that is invariant to the

We compare our method to [ 17 ], where the authors learn a set of Mahalanobis-like metrics independently from each other: they cluster the color patches using k-means and learn a

Such sets we called simultaneous metric generators in [ 22 ], where, by analogy, a simultaneous metric basis was defined as a simultaneous metric generator of minimum cardinality,

The predictive model M defines a conditional probability distribution of target vari- ables given the value of context variables: p(A|C) and it is induced (learned) from examples in

Finally, the package implements one triplet learner and one quadruplet learner: Sparse Compositional Metric Learning (SCML, Shi et al., 2014) and Metric Learning from

Unlike the latter, the barycentric representation allows us to work directly on Euclidean space and apply a metric learning algorithm to find a suitable metric that is

2.3 Conflict Analysis and Implication Graph Now, we introduce a fundamental data structure, called implication graph, used by complete CDCL solvers (Con- flict Driven Clause