Predicting Structured Data (Neural Information Processing)

By Gükhan H. Bakir, Thomas Hofmann, Bernhard Schölkopf, Alex Smola, B. Taskar, S. V. N. Vishwanathan. 2007

136 citations recorded by the source corpus.

How researchers see this paper

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

In Information Extraction

Citations in the corpus

Value Iteration Networks (2016, methodology): “…most recent deep RL works [19, 16, 10] employed NN architectures that are very similar to the standard networks used in supervised learning tasks, which typically consist of CNNs for feature extraction, and fully connected layers that map the features to a probability distribution over actions.”
An Algorithmic Perspective on Imitation Learning (2018, background): “The task of learning a function that maps inputs x to structured outputs y (for example, parse trees, trajectories, matchings, etc. [Taskar, 2005]) is referred to as structured prediction [Tsochantaridis et al., 2005, BakIr et al., 2007].”
Hybrid Markov Logic Networks (2008, background): “This problem has received much attention in recent years, and many approaches have been proposed (Bakir et al. 2007; Getoor & Taskar 2007).”
Unsupervised Ontology Induction from Text (2010, background): “Combining uncertainty handling and joint inference is the hallmark of the emerging field of statistical relational learning (a.k.a. structured prediction), where a plethora of approaches have been developed (Getoor and Taskar, 2007; Bakir et al., 2007).”
Soft-DTW: a Differentiable Loss Function for Time-Series (2017, background): “This task is noticeably more difficult when the output objects have a structure, i.e. when they are not vectors (Bakir et al., 2007).”
Learning Output Kernels with Block Coordinate Descent (2011, methodology): “…by fixing a basis for the output space, the problem of learning a vector-valued function can be equivalently regarded as a problem of learning a scalar function defined over an enlarged input set, a modeling approach that is also used in the structured output learning setting (Bakir et al., 2007).”
Structured Prediction Theory Based on Factor Graph Complexity (2016, result): “However, in contrast to the plethora of algorithms, there have been relatively few studies devoted to the theoretical understanding of structured prediction problems [Bakir et al., 2007].”
Machine Reading: A "Killer App" for Statistical Relational AI (2010, background): “Recent research in statistical relational AI offers a plethora of approaches for joint inference and learning (Getoor and Taskar 2007; Bakir et al. 2007).”
A Generalized Kernel Approach to Structured Output Learning (2012, background): “Two different, but closely related, kernel-based approaches for structured output learning can be found in the literature (Bakir et al., 2007): kernel dependency estimation (KDE) and joint kernel maps (JKM).”
Hinge-Loss Markov Random Fields and Probabilistic Soft Logic (2015, background): “A related field to SRL is structured prediction (SP) (Bakir et al., 2007; Nowozin et al., 2016), which generalizes the tasks of classification and regression to the task of predicting structured objects.”
Learning to Connect Language and Perception (2008, background): “If given sentences with sets of possible meaning representations, it constitutes a form of MIL for learning with structured data (Bakir et al. 2007).”
Unsupervised Semantic Parsing (2009, background): “In many NLP applications, there exist rich relations among objects, and recent work in statistical relational learning (Getoor and Taskar, 2007) and structured prediction (Bakir et al., 2007) has shown that leveraging these can greatly improve accuracy.”
Machine Reading at the University of Washington (2010, background): “Past work has shown that it can substantially improve predictive accuracy in supervised learning (e.g., (Getoor and Taskar, 2007; Bakir et al., 2007)).”
Deep Neural Network for Structural Prediction and Lane Detection in Traffic Scene (2017, background): “However, in more challenging scenarios, such as detecting objects of varying sizes and shapes, making detections from the simple outputs requires sophisticated structuring and is nontrivial [5].”