Approximate Cross-Validation for Structured Models
This repository contains code for approximate cross-validation for structred models based on the infinitesimal jackknife as developed in .
conda env create -f environment.yml conda activate approxcv
To perform LWCV on a AR(0) hidden Markov model trained on synthetic data,
To perform LWCV on a discrete MRF,
 Ghosh, Soumya*, William T. Stephenson*, Tin D. Nguyen, Sameer K. Deshpande, and Tamara Broderick. Approximate Cross-Validation for Structured Models. NeurIPS 2020.
* Equal contribution