SVRG and Beyond via Posterior Correction
By: Nico Daheim , Thomas Möllenhoff , Ming Liang Ang and more
Potential Business Impact:
Makes AI learn faster and better.
Stochastic Variance Reduced Gradient (SVRG) and its variants aim to speed-up training by using gradient corrections, but have seen limited success in deep learning. Here, we show surprising new foundational connections of SVRG to a recently proposed Bayesian method called posterior correction. Specifically, we show that SVRG is recovered as a special case of posterior correction over the isotropic-Gaussian family, while novel extensions are automatically obtained by using more flexible exponential families. We derive two new SVRG variants by using Gaussian families: First, a Newton-like variant that employs novel Hessian corrections, and second, an Adam-like extension that improves pretraining and finetuning of Transformer language models. This is the first work to connect SVRG to Bayes and use it to boost variational training for deep networks.
Similar Papers
On the convergence of stochastic variance reduced gradient for linear inverse problems
Numerical Analysis
Solves hard math problems faster and more accurately.
Convergence Analysis of alpha-SVRG under Strong Convexity
Machine Learning (CS)
Makes computer learning faster and better.
Uncertainty Quantification for Scientific Machine Learning using Sparse Variational Gaussian Process Kolmogorov-Arnold Networks (SVGP KAN)
Machine Learning (CS)
Shows how sure a computer is about its answers.