DIFFRAC: a discriminative and flexible framework for clustering

By Francis R. Bach, Zaïd Harchaoui. 2007

137 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning, Probabilistic & Bayesian Methods

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

In Probabilistic & Bayesian Methods

Citations in the corpus

End-to-End Learning of Visual Representations From Uncurated Instructional Videos (2019, background): “DIFFRAC [10]) to resolve the label ambiguities, and have used mostly linear (or shallow) models.”
Joint Discovery of Object States and Manipulation Actions (2017, methodology): “In particular, we build on the discriminative clustering approach of [4] that has been shown to perform well in a variety of computer vision problems [5].”
Unsupervised Learning from Narrated Instruction Videos (2015, background): “The values of precision and recall are given at the bottom of Table 1.”
Deep Clustering for Unsupervised Learning of Visual Features (2018, background): “Solutions are typically based on constraining or penalizing the minimal number of points per cluster [2, 26].”
Unsupervised Learning by Predicting Noise (2017, methodology): “Finally, we provide all the implementation details.”
Weakly Supervised Action Labeling in Videos under Ordering Constraints (2014, methodology): “Yet another way to evaluate is to use a loss between partitions [14] as in [3].”
Self-labelling via simultaneous clustering and representation learning (2019, background): “Finally, (Bach & Harchaoui, 2008; Vo et al., 2019) use convex relaxations to regularized afﬁne-transformation invariant linear clustering, that does not scale to larger datasets.”
Cross-Task Weakly Supervised Learning From Instructional Videos (2019, methodology): “, minimizing the L2 loss with linear model [3, 2]) or an unrestricted classifier, such as a deep network, but no constraints [4, 7].”
Weakly-Supervised Alignment of Video with Text (2015, methodology): “Contrary to standard clustering techniques, it uses a discriminative cost function [2, 16] and it has been used in image co-segmentation [20, 21], object colocalization [22], person identification in video [6, 36], and alignment of labels to videos [7].”
Weakly-Supervised Learning of Visual Relations (2017, methodology): “Our work uses a discriminative clustering objective [3], that has been successful in several computer vision tasks [5, 24], but has not been so far, to the best of our knowledge, used for modeling relations.”
Learning from Video and Text via Large-Scale Discriminative Clustering (2017, methodology): “The objective function f of problem (3) is f(Y ) = 〈Y,A(X,λ)Y 〉F, where (see [2])”
Multi-class cosegmentation (2012, background): “A common solution is to add constraints on the number of elements per class [10, 15].”
Finding Actors and Actions in Movies (2013, methodology): “Following [2], we note that (2) admits a closed form solution in w and b for fixed Z.”
Instance-Level Video Segmentation from Object Tracks (2016, methodology): “The second term is a discriminative clustering cost based on the square loss [3] which learns a foreground vs.”
Trace optimization and eigenproblems in dimension reduction methods (2011, unknown): “, [57, 58, 3].”
Maximum Margin Temporal Clustering (2012, methodology): “In contrast, we propose to incorporate discriminative clustering, which has been shown to have advantages over generative models (Bach and Harchaoui, 2009; De la Torre and Kanade, 2006; Zhao et al., 2008), into the TC problem.”
Unsupervised Image Matching and Object Discovery as Optimization (2019, background): “2Note that plain unsupervised clustering, whether classic, spectral, discriminative or deep [4, 22, 32, 36], focuses on data partitioning and not on the discovery of subsets of matching items within a cluttered collection.”
Convex Discriminative Multitask Clustering (2013, methodology): “This phenomenon has been observed in both the supervised classification [26] and the clustering [27].”
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].”
Discriminative Clustering by Regularized Information Maximization (2010, methodology): “Like our approach, the works of [2] and [21] use notions of class balance, seperation, and regularization to learn unsupervised discriminative classifiers.”