Infinite-dimensional gradient-based descent for alpha-divergence minimisation

Abstract

This paper introduces the $(\alpha ,\Gamma )$-descent, an iterative algorithm which operates on measures and performs $\alpha$-divergence minimisation in a Bayesian framework. This gradient-based procedure extends the commonly-used variational approximation by adding a prior on the variational parameters in the form of a measure. We prove that for a rich family of functions $\Gamma$, this algorithm leads at each step to a systematic decrease in the $\alpha$-divergence and derive convergence results. Our framework recovers the Entropic Mirror Descent algorithm and provides an alternative algorithm that we call the Power Descent. Moreover, in its stochastic formulation, the $(\alpha ,\Gamma )$-descent allows to optimise the mixture weights of any given mixture model without any information on the underlying distribution of the variational parameters. This renders our method compatible with many choices of parameters updates and applicable to a wide range of Machine Learning tasks. We demonstrate empirically on both toy and real-world examples the benefit of using the Power descent and going beyond the Entropic Mirror Descent framework, which fails as the dimension grows.

References in corpus (10)

• Expectation Propagation for approximate Bayesian inference
• Variational Bayesian Inference with Stochastic Search
• Nonparametric variational inference
• On $f$-Divergences: Integral Representations, Local Behavior, and Inequalities
• Semi-Implicit Variational Inference
• Variational Inference with Tail-adaptive f-Divergence
• Finding Mixed Nash Equilibria of Generative Adversarial Networks
• A Generalization Bound for Online Variational Inference
• Perturbative Black Box Variational Inference
• Safe and adaptive importance sampling: a mixture approach

Cited by in corpus (1)

• Adaptive Importance Sampling meets Mirror Descent: a Bias-variance tradeoff