Maximum Margin Clustering

By Linli Xu, James Neufeld, Bryce Larson, Dale Schuurmans. 2004

528 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

Joint Discovery of Object States and Manipulation Actions (2017, methodology): “Our model builds on unsupervised discriminative clustering methods [4, 41, 46] that group data samples according to a simultaneously learned classifier.”
Deep Clustering for Unsupervised Learning of Visual Features (2018, background): “An optimal decision boundary is to assign all of the inputs to a single cluster [67].”
Unsupervised Learning by Predicting Noise (2017, methodology): “As opposed to Xu et al. (2004), we address the empty assignment problems by restricting the set of possible reassignments to permutations rather than using global linear constrains the assignments.”
Weakly Supervised Action Labeling in Videos under Ordering Constraints (2014, methodology): “Following recent work on discriminative clustering [3,37], image [18] and video [5] cosegmentation, we propose to exploit this information in a discriminative framework where both the action model and the optimal assignments under temporal constraints are learned together.”
Cross-Task Weakly Supervised Learning From Instructional Videos (2019, background): “Our technical approach to the problem builds particularly heavily on the use of discriminative clustering [3, 32], or the simultaneous constrained grouping of data samples and learning of classifiers for groups.”
Stochastic Variance Reduction Methods for Saddle-Point Problems (2016, background): “g, for discriminative clustering [17] or matrix factorization [18].”
Learning Discrete Representations via Information Maximizing Self-Augmented Training (2017, methodology): “…methods for clustering and hash learning can be broadly divided into two categories according to the decision boundaries they can model: linear (He et al., 2013; Gong et al., 2013; Norouzi & Blei, 2011) and non-linear (Xu et al., 2004; Kulis & Darrell, 2009; Ng et al., 2001; Weiss et al., 2009).”
Convex Multi-Task Learning by Clustering (2015, background): “The final formulation is a saddle-point convex problem in both α and E, which follows similar arguments applied in other contexts [17, 25, 22].”
Trading convexity for scalability (2006, methodology): “Other methods [2, 22] transform the non-convex transductive problem into a convex semi-definite programming that scales as (L+ U)(4) or worse.”
Unsupervised Supervised Learning II: Margin-Based Classification without Labels (2011, unknown): “Related problems have been addressed in [13, 10, 5].”
Co-localization in Real-World Images (2014, methodology): “We can take advantage of these methods even in our unsupervised scenario, where we do not know the labels of our boxes [2, 34].”
Automatic annotation of human actions in video (2009, methodology): “In this section, we formulate the above clustering problem as a minimization of a discriminative cost function [23].”
A Comprehensive Survey of Clustering Algorithms (2015, methodology): “MKC, an improvement of MMC, tries to find the best hyperplane based on several kernels to cluster.”
Deep Clustering via Joint Convolutional Autoencoder Embedding and Relative Entropy Minimization (2017, background): “Information theoretic [18, 2, 14], maxmargin [64, 55], and spectral graph [24] algorithms are examples of discriminative clustering models.”
Support Vector Machines as Probabilistic Models (2011, background): “The Maximum Margin Clustering (MMC) (Xu et al., 2004) is a popular heuristic that transfers the maximum margin principle from supervised SVM learning to the unsupervised setting.”
Large Scale Transductive SVMs (2006, methodology): “Other methods (Bie and Cristianini, 2004; Xu et al., 2005) transform the non-convex transductive problem into a convex semi-deﬁnite programming problem that scales as ( L + U ) 4 or worse.”
Optimization Techniques for Semi-Supervised Support Vector Machines (2008, methodology): “We follow Bie and Cristianini (2006) in this section, but outline the details for the squared Hinge loss (see also Xu et al., 2004, for a similar derivation).”
Multiple Kernel Fuzzy Clustering (2012, background): “Over the past decades, many clustering algorithms have been proposed, including k-means clustering [1], mixture models [1], spectral clustering [2], localitysensitive hashing [3], and maximum margin clustering [4], [5].”
$\propto$SVM for learning with label proportions (2013, methodology): “ing, 2010; Quadrianto et al., 2009), /SVM requires no restrictive assumptions on the data. In fact, in the special case where no label proportions are provided, /SVM becomes large-margin clustering (Xu et al., 2004; Li et al., 2009a), whose solution depends only on the data distribution. /SVM can na”