Corpus ID: 16404669

Exact Sampling with Integer Linear Programs and Random Perturbations

@inproceedings{Kim2016ExactSW,
  title={Exact Sampling with Integer Linear Programs and Random Perturbations},
  author={Carolyn Kim and Ashish Sabharwal and S. Ermon},
  booktitle={AAAI},
  year={2016}
}
We consider the problem of sampling from a discrete probability distribution specified by a graphical model. [...] Key Method Our technique, GumbelMIP, leverages linear programming (LP) relaxations to evaluate the quality of samples and prune large portions of the search space, and can thus scale to large tree-width models beyond the reach of current exact inference methods. Further, when the optimization problem is not solved to optimality, our method yields a novel approximate sampling technique. We…Expand
High Dimensional Inference With Random Maximum A-Posteriori Perturbations
Direct Policy Gradients: Direct Optimization of Policies in Discrete Action Spaces
Approximate Inference via Weighted Rademacher Complexity
Lost Relatives of the Gumbel Trick
Estimating Gradients for Discrete Random Variables by Sampling without Replacement
Fast Generating A Large Number of Gumbel-Max Variables
Learning and Inference via Maximum Inner Product Search
...
1
2
...

References

SHOWING 1-10 OF 29 REFERENCES
A* Sampling
Randomized Optimum Models for Structured Prediction
Perturb-and-MAP random fields: Using discrete optimization to learn and sample from energy models
Approximation Algorithms for NP-Hard Problems
Progress in Linear Programming-Based Algorithms for Integer Programming: An Exposition
Approximate Counting by Sampling the Backtrack-free Search Space
On Sampling from the Gibbs Distribution with Random Maximum A-Posteriori Perturbations
Graphical Models, Exponential Families, and Variational Inference
The OS* Algorithm: a Joint Approach to Exact Optimization and Sampling
SampleSearch: Importance sampling in presence of determinism
...
1
2
3
...