Distance Metric Learning with Application to Clustering with Side-Information

By Eric P. Xing, Andrew Y. Ng, Michael I. Jordan, Stuart J. Russell. 2002

3,352 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

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

In Representation & Self-Supervised Learning

Citations in the corpus

Unsupervised state representation learning with robotic priors: a robustness benchmark (2017, methodology): “…our approach, the priors are used as loss functions that encode constraints between states, a conﬁguration that we address using Siamese networks (e.g., [16], [17], [6]), which use two (or more) copies of a network with tied weights to process two (or more) inputs whose relation has to be imposed.”
Deep Weighted Averaging Classifiers (2018, methodology): “This is closely related to a long line of past work on metric learning [3, 10, 23, 58, 60], but rather than trying to optimize a particular notion of distance (such as Mahalanobis distance), we make use of a fixed distance function, and allow the architecture of standard deep models to do the equiva”
Word Representations via Gaussian Embedding (2014, background): “Other work in multiplicative tensor factorization for word embeddings (Kiros et al., 2014) and metric learning (Xing et al., 2002) learns some combinations of representations, clusters, and a distance metric jointly; however, it does not effectively learn a distance function per item.”
Learning to cluster in order to Transfer across domains and tasks (2017, background): “For example, DML (Xing et al., 2003), ITML (Davis et al.”
SphereFace: Deep Hypersphere Embedding for Face Recognition (2017, background): “Traditional metric learning [36, 33, 12, 38] usually learns a matrix A for a distance metric ‖x1−x2‖A= √ (x1−x2)A(x1−x2) upon the given features x1,x2.”
Unsupervised Deep Embedding for Clustering Analysis (2015, background): “…studied in machine learning in terms of feature selection (Boutsidis et al., 2009; Liu & Yu, 2005; Alelyani et al., 2013), distance functions (Xing et al., 2002; Xiang et al., 2008), grouping methods (Mac-Queen et al., 1967; Von Luxburg, 2007; Li et al., 2004), and cluster validation…”
Learning Invariant Feature Spaces to Transfer Skills with Reinforcement Learning (2017, methodology): “Xing et al. (2002) ﬁnds a linear transformation of the input data to satisfy pairwise similarity contraints, while past work by Chopra et al. (2005) used Siamese networks to learn a feature space where paired images are brought close together and unpaired images are pushed apart.”
Multi-modal Transformer for Video Retrieval (2020, background): “start", it needs to use cross-modal information (e.g., absence of noise after someone tried to start the engine, orange arrow). A common approach for the retrieval problem is similarity learning [29], where we learn a function of two elements (a query and a candidate) that best describes their ”
Sampling Matters in Deep Embedding Learning (2017, neutral): “Convex approaches were much more popular [7, 39].”
Petuum: A New Platform for Distributed Machine Learning on Big Data (2013, background): “DML improves the performance of other ML programs such as clustering, by allowing domain experts to incorporate prior knowledge of the form “data points x , y are similar (or dissimilar)” [29] — for example, we could enforce that “books about science are diﬀerent from books about art”.”
A Survey on Metric Learning for Feature Vectors and Structured Data (2013, background): “For metric learning from structured data, we assume that the reader has some familiarity with basic probability theory, statistics and likelihood maximization.”
Learning Distance Metrics with Contextual Constraints for Image Retrieval (2006, methodology): “…methods are compared as follows: (1) Euclidean: retrieval by Euclidean metric; (2) RCA: retrieval by the metric of RCA; (3) Xing: retrieval by the metric of Xing’s method with nonlinear optimization [21]; (4) DCA: retrieval by the metric of DCA; (5) KDCA: retrieval by the nonlinear metric of KDCA.”
Learning a Nonlinear Embedding by Preserving Class Neighbourhood Structure (2007, neutral): “Comparison to other recent methods for learning similarity measures [1, 10, 20] remains to be done.”
Retrieval of Brain Tumors by Adaptive Spatial Pooling and Fisher Vector Representation (2016, background): “Optimization with respect toM often needs the projection ofM to the PSD cone by setting the negative eigenvalues to zero at each iteration [30,33].”
Semi-supervised Kernel Metric Learning Using Relative Comparisons (2016, background): “Most of the metric-learning literature, starting by the work of Xing et al. [2002], aims at ﬁnding a Mahalanobis matrix subject to either ML/CL or relative distance constraints.”
Local Modeling of Attributed Graphs: Algorithms and Applications (2016, positive): “We adopt the optimization objective by [148]:”