Corpus ID: 8548126

Bethe Projections for Non-Local Inference

@inproceedings{Vilnis2015BethePF,
  title={Bethe Projections for Non-Local Inference},
  author={L. Vilnis and David Belanger and D. Sheldon and A. McCallum},
  booktitle={UAI},
  year={2015}
}
Many inference problems in structured prediction are naturally solved by augmenting a tractable dependency structure with complex, non-local auxiliary objectives. This includes the mean field family of variational inference algorithms, soft- or hard-constrained inference using Lagrangian relaxation or linear programming, collective graphical models, and forms of semi-supervised learning such as posterior regularization. We present a method to discriminatively learn broad families of inference… Expand
16 Citations
Inference with Aggregate Data: An Optimal Transport Approach
  • 7
  • Highly Influenced
  • PDF
Building More Expressive Structured Models
  • 1
  • PDF
Graph Structured Prediction Energy Networks
  • 6
  • PDF
Cooperative Graphical Models
  • 1
  • PDF
End-to-End Learning for Structured Prediction Energy Networks
  • 92
  • PDF
Learning Hidden Markov Models from Aggregate Observations
  • Highly Influenced
  • PDF
Differentially Private Learning of Undirected Graphical Models Using Collective Graphical Models
  • 18
  • PDF
Graphical-model based estimation and inference for differential privacy
  • 17
  • PDF
...
1
2
...

References

SHOWING 1-10 OF 46 REFERENCES
Posterior Regularization for Structured Latent Variable Models
  • 455
  • PDF
Approximate Inference in Collective Graphical Models
  • 30
  • PDF
Graphical Models, Exponential Families, and Variational Inference
  • 3,430
  • PDF
1 Introduction to Dual Decomposition for Inference
  • 180
  • PDF
Structured Output Learning with High Order Loss Functions
  • 55
  • PDF
Learning Soft Linear Constraints with Application to Citation Field Extraction
  • 15
  • PDF
Max-Margin Markov Networks
  • 1,461
  • PDF
Learning from measurements in exponential families
  • 117
  • PDF
...
1
2
3
4
5
...