Learning structured prediction models: a large margin approach

By B. Taskar, Vassil Chatalbashev, Daphne Koller, Carlos Guestrin. 2005

575 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, 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 Computer Vision and Pattern Recognition

In Natural Language Processing

In Optimization & Training

Citations in the corpus

On Structured Prediction Theory with Calibrated Convex Surrogate Losses (2017, background): “It is known that Crammer-Singer multi-class SVM [14], which SSVM is built on, is not consistent for 0-1 loss unless there is a majority class with probability at least [45, 28].”
Minding the Gaps for Block Frank-Wolfe Optimization of Structured SVMs (2016, background): “The parameter ν controls the global component and acts as a safety parameter to ensure convergence (Theorem 3).”
Block-Coordinate Frank-Wolfe Optimization for Structural SVMs (2012, methodology): “Interestingly, for structural SVMs we also show that existing batch subgradient and cutting-plane methods are special cases of Frank-Wolfe algorithms, and this leads to stronger and simpler O(1/ε) convergence rate guarantees for these existing algorithms. ar X iv :1 20 7.”
Fast Edge Detection Using Structured Forests (2014, background): “Our novel approach to learning decision trees uses structured labels to determine the splitting function at each branch in the tree.”
Search-based structured prediction (2009, positive): “The maximum margin Markov network (M 3 N) formalism considers the structured prediction problem as a quadratic programming problem [52,51], following the formalism for the support vector machine for binary classiﬁcation.”
Learning Deep Structured Models (2014, methodology): “In the future we plan to learn deeper models for object recognition in applications such as holistic semantic scene understanding.”
OptNet: Differentiable Optimization as a Layer in Neural Networks (2017, methodology): “There are two broad classes of learning approaches in structured prediction: methods that use probabilistic inference techniques (typically exploiting the fact that the gradient of the log likelihood is given by the actual feature expectations minus their expectation under the learned model (Koller ”
An Algorithmic Perspective on Imitation Learning (2018, background): “…is closely related to supervised learning, and is particularly related to the field of structured prediction [Daumé III et al., 2009, Ratliff et al., 2006a, Taskar, 2005] , where the task is to learn a mapping from inputs x to a complex, structured output y (plans, parse trees, complex motions).”
Input Convex Neural Networks (2016, methodology): “Although max-margin structured prediction is a simple and well-studied approach (Tsochantaridis et al., 2005; Taskar et al., 2005), in our experiences using these methods within an ICNN, we had substantial difﬁculty choosing the proper margin scaling term (especially for domains with…”
Simple and Accurate Dependency Parsing Using Bidirectional LSTM Feature Representations (2016, methodology): “In order to remedy this, we found it useful to use loss augmented inference (Taskar et al., 2005).”
Imitation learning for locomotion and manipulation (2007, methodology): “In what follows, we review the structured margin multiclass classification setting of [3] used throughout this paper which explicitly utilizes information of this sort.”
Learn to Solve Algebra Word Problems Using Quadratic Programming (2015, methodology): “Motivated by the work (Taskar et al., 2005; Li, 2014), we adopt the max-margin objective.”
Safe, Multi-Agent, Reinforcement Learning for Autonomous Driving (2016, background): “e horizon and thereby reducing the variance of the gradient estimation even further (detailed in sec. 5). The Option Graph plays a similar role to “structured prediction” in supervised learning (e.g. [36]), thereby reducing sample complexity, while also playing a similar role to LSTM [14] gating mec”
Connectionist multivariate density-estimation and its application to speech synthesis (2016, background): “Structured prediction (Taskar, 2004), which is closely related to unsupervised learning, models the probability distribution among the dependant variables of a conditional dimensions conditioned on the value of other dimensions is only tractable in the few particular cases where the dimensions…”
Link-Based Classification (2003, methodology): “[21] used the link-based classification formulation for link prediction to predict disulphide bonds in proteins.”
Edge-Labeling Graph Neural Network for Few-Shot Learning (2019, background): “Taskar [42] derived a maxmargin formulation for learning the edge scores in CC for producing two different segmentations of a single image.”
Distributed Asynchronous Online Learning for Natural Language Processing (2010, background): “Another example is large-margin learning for structured prediction (Taskar et al., 2005; Tsochan- 1We use “convex” to mean convex-up when minimizing and convex-down, or concave, when maximizing. taridis et al., 2005), which can be solved by subgradient methods (Ratliff et al., 2006).”
Boosting Structured Prediction for Imitation Learning (2006, methodology): “Unfortunately, this Maximum Margin Planning (MMP) approach, as well as related techniques for maximum margin structured learning developed in (Taskar et al., 2005) and (Taskar et al.”