Structured Prediction, Dual Extragradient and Bregman Projections

By B. Taskar, Simon Lacoste-Julien, Michael I. Jordan. 2006

130 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Optimization & Training

How researchers see this paper

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

In Artificial Intelligence

In Optimization & Training

Citations in the corpus

Frank-Wolfe Algorithms for Saddle Point Problems (2016, methodology): “We consider the saddle point formulation (Taskar et al., 2006) for a `1-regularized structured SVM objective that minimizes the primal cost function p(w) := 1 n ∑n i=1 H̃i(w), where H̃i(w) = maxy∈Yi Li(y)−〈w,ψi(y)〉 is the structured hinge loss (using the notation from Lacoste-Julien et al.”
Block-Coordinate Frank-Wolfe Optimization for Structural SVMs (2012, methodology): “For testing, error is the ‘alignment error rate’, as defined in Taskar et al. (2006).”
Stochastic Variance Reduction Methods for Saddle-Point Problems (2016, background): “Following ∈ In this situation, prox σM ( x ′ , y ′ ) = (prox σf ( x ′ ) , prox σg ( y ′ )) the usual set-up of composite optimization [10], no smoothness assumption is made on M and hence on f or g .”
Hidden Part Models for Human Action Recognition: Probabilistic versus Max Margin (2011, methodology): “It has been shown that if E forms a forest, the optimal solution of this LP is guaranteed to be integral [29], [39].”
Learning Mixtures of Submodular Shells with Application to Document Summarization (2012, methodology): “Many algorithms have been proposed for the large margin learning problem, including those based on the exponentiated gradient method [7], the dual extragadient method [52], the cutting-plane algorithm [54], and the subgradient descent method [42, 43].”
Accelerated training of max-margin Markov networks with kernels (2011, background): “Extragradient [10] primal-dual duality gap exp n/a O ( log|Y| )”
Structured Prediction via the Extragradient Method (2005, methodology): “In [ 21 ], we show how to apply this approach efficiently to other types of models, includin g general Markov networks and weighted context-free grammars, using Bregman projections.”