Titlebook: Metric Learning; Aurélien Bellet,Amaury Habrard,Marc Sebban Book 2015 Springer Nature Switzerland AG 2015

Ambiguous

歌剧等

Introduction,ining techniques. Prominent examples include nearest-neighbor classification [Cover and Hart, 1967], data clustering [Lloyd, 1982], kernel methods [Schölkopf and Smola, 2001] and many information retrieval methods [Manning et al., 2008].

peritonitis

Metric Learning for Structured Data,d Perona, 2005, Salton et al., 1975]. In this case, metric learning can simply be performed on the feature vector representation, but this strategy can imply a significant loss of structural information.

Morphine

刺激

幻想

Book 2015 appropriately measure such similarities for a given task is crucial to the performance of many machine learning, pattern recognition and data mining methods. This book is devoted to metric learning, a set of techniques to automatically learn similarity and distance functions from data that has attr

judicial

Linear Metric Learning, devoted to learning more flexible linear similarity functions (Section 4.2). Finally, Section 4.3 discusses how to scale-up these methods to large amounts of training data (both in number of samples and number of features).

giggle

Nonlinear and Local Metric Learning,ce. In this chapter, we present the two main lines of research in nonlinear metric learning: learn a nonlinear form of metric (Section 5.1) or multiple linear metrics (Section 5.2), as illustrated in Figure 5.1.

A保存的

Generalization Guarantees for Metric Learning,on the training sample). This deviation is typically a function of the number of training examples, and some notion of complexity of the model such as the VC dimension [Vapnik and Chervonenkis, 1971], the fat-shattering dimension [Alon et al., 1997] or the Rademacher complexity [Bartlett and Mendelson, 2002, Koltchinskii, 2001].

完成