Surrogate Likelihood

Surrogate likelihood methods aim to replace computationally expensive or intractable likelihood functions in statistical inference, enabling efficient Bayesian inference and model training, particularly with large datasets. Current research focuses on developing surrogate likelihoods using neural networks, normalizing flows, and other flexible models, often within distributed or communication-efficient algorithms for high-dimensional data. These techniques are significantly impacting various fields, improving the scalability and practicality of Bayesian inference in applications ranging from astrophysics and neuroscience to large-scale machine learning. The development of theoretical guarantees for the accuracy and consistency of these surrogates remains an active area of investigation.

Papers

August 30, 2024

Minimax and Communication-Efficient Distributed Best Subset Selection with Oracle Property
Jingguo Lan, Hongmei Lin, Xueqin Wang
Test Oracle Minimax Optimality Sparse Regression Sparsity Pattern Best Subset Surrogate Likelihood Sparse Estimation

July 29, 2024

Neural Surrogate HMC: Accelerated Hamiltonian Monte Carlo with a Neural Network Surrogate Likelihood
Linnea M Wolniewicz, Peter Sadowski, Claudio Corti
Bayesian Inference Markov Chain Monte Carlo Hamiltonian Monte Carlo Log Likelihood Likelihood Based Neural Surrogate Surrogate Likelihood

July 8, 2024

High-Dimensional Distributed Sparse Classification with Scalable Communication-Efficient Global Updates
Fred Lu, Ryan R. Curtin, Edward Raff, Francis Ferraro, James Holt
High Dimension Logistic Regression Statistical Learning Linear Model Parameter Update Sparse Classification Surrogate Likelihood

May 20, 2024

General bounds on the quality of Bayesian coresets
Trevor Campbell
Quality Issue General Text COreset Selection Asymptotic Inference Surrogate Likelihood

August 2, 2023

Simulation-based inference using surjective sequential neural likelihood estimation
Simon Dirmeier, Carlo Albert, Fernando Perez-Cruz
Simulation Based Inference Likelihood Based Surrogate Likelihood Sequential Neural

February 10, 2023

Efficient and Accurate Learning of Mixtures of Plackett-Luce Models
Duc Nguyen, Anderson Y. Zhang
High Efficiency Mixture Component Maximum Likelihood Log Likelihood LLM Model EM Algorithm Accurate Learning Surrogate Likelihood

October 21, 2022

Efficient identification of informative features in simulation-based inference
Jonas Beck, Michael Deistler, Yves Bernaerts, Jakob Macke, Philipp Berens
Simulation Based Inference Identification Accuracy Surrogate Likelihood Posterior Uncertainty Neural Likelihood Hodgkin Huxley

June 18, 2022

The Consistency of Adversarial Training for Binary Classification
Natalie S. Frank, Jonathan Niles-Weed
Adversarial Training Adversarial Perturbation Strong Consistency Binary Classification Adversarial Risk Adversarial Setting Surrogate Likelihood Adversarial Surrogate

December 22, 2021

Surrogate Likelihoods for Variational Annealed Importance Sampling
Martin Jankowiak, Du Phan
Variational Inference Hamiltonian Monte Carlo Approximate Bayesian Inference Annealed Importance Sampling Variational Parameter Surrogate Likelihood