Autoregressive Models: What Are They Good For?

10/17/2019

∙

by   Murtaza Dalal, et al.

∙

37

∙

share

Autoregressive (AR) models have become a popular tool for unsupervised learning, achieving state-of-the-art log likelihood estimates. We investigate the use of AR models as density estimators in two settings – as a learning signal for image translation, and as an outlier detector – and find that these density estimates are much less reliable than previously thought. We examine the underlying optimization issues from both an empirical and theoretical perspective, and provide a toy example that illustrates the problem. Overwhelmingly, we find that density estimates do not correlate with perceptual quality and are unhelpful for downstream tasks.