Preferential Subsampling for Stochastic Gradient Langevin Dynamics
This addresses variance control in scalable Bayesian inference for practitioners, though it is incremental as it builds on existing SGMCMC methods.
The paper tackles the high variance problem in stochastic gradient MCMC by introducing preferential subsampling and adaptive subsample size adjustment, resulting in maintaining accuracy while substantially reducing the average subsample size.
Stochastic gradient MCMC (SGMCMC) offers a scalable alternative to traditional MCMC, by constructing an unbiased estimate of the gradient of the log-posterior with a small, uniformly-weighted subsample of the data. While efficient to compute, the resulting gradient estimator may exhibit a high variance and impact sampler performance. The problem of variance control has been traditionally addressed by constructing a better stochastic gradient estimator, often using control variates. We propose to use a discrete, non-uniform probability distribution to preferentially subsample data points that have a greater impact on the stochastic gradient. In addition, we present a method of adaptively adjusting the subsample size at each iteration of the algorithm, so that we increase the subsample size in areas of the sample space where the gradient is harder to estimate. We demonstrate that such an approach can maintain the same level of accuracy while substantially reducing the average subsample size that is used.