Stochastic Gradient MCMC with Repulsive Forces
arXiv:1812.00071
Abstract
We propose a unifying view of two different Bayesian inference algorithms, Stochastic Gradient Markov Chain Monte Carlo (SG-MCMC) and Stein Variational Gradient Descent (SVGD), leading to improved and efficient novel sampling schemes. We show that SVGD combined with a noise term can be framed as a multiple chain SG-MCMC method. Instead of treating each parallel chain independently from others, our proposed algorithm implements a repulsive force between particles, avoiding collapse and facilitating a better exploration of the parameter space. We also show how the addition of this noise term is necessary to obtain a valid SG-MCMC sampler, a significant difference with SVGD. Experiments with both synthetic distributions and real datasets illustrate the benefits of the proposed scheme.
arXiv commentExtends the workshop version
References in corpus (8)
- Stochastic Gradient Hamiltonian Monte Carlo
- Preconditioned Stochastic Gradient Langevin Dynamics for Deep Neural Networks
- Cyclical Stochastic Gradient MCMC for Bayesian Deep Learning
- Stein Variational Gradient Descent as Gradient Flow
- Yes, but Did It Work?: Evaluating Variational Inference
- Bridging the Gap between Stochastic Gradient MCMC and Stochastic Optimization
- A Unified Particle-Optimization Framework for Scalable Bayesian Sampling
- Meta-Learning for Stochastic Gradient MCMC
Cited by in corpus (12)
- 3D Bayesian Variational Full Waveform Inversion
- On the geometry of Stein variational gradient descent
- Adversarial Machine Learning: Bayesian Perspectives
- Annealed Stein Variational Gradient Descent for Improved Uncertainty Estimation in Full-Waveform Inversion
- Protecting Classifiers From Attacks
- On Stein Variational Neural Network Ensembles
- Stein Self-Repulsive Dynamics: Benefits From Past Samples
- De-randomizing MCMC dynamics with the diffusion Stein operator
- Understanding MCMC Dynamics as Flows on the Wasserstein Space
- Variationally Inferred Sampling Through a Refined Bound for Probabilistic Programs
- Neural Variational Gradient Descent
- Contributions to Large Scale Bayesian Inference and Adversarial Machine Learning