Natural Image Manipulation for Autoregressive Models Using Fisher Scores [article]

Wilson Yan, Jonathan Ho, Pieter Abbeel
2020 arXiv   pre-print
Deep autoregressive models are one of the most powerful models that exist today which achieve state-of-the-art bits per dim. However, they lie at a strict disadvantage when it comes to controlled sample generation compared to latent variable models. Latent variable models such as VAEs and normalizing flows allow meaningful semantic manipulations in latent space, which autoregressive models do not have. In this paper, we propose using Fisher scores as a method to extract embeddings from an
more » ... gressive model to use for interpolation and show that our method provides more meaningful sample manipulation compared to alternate embeddings such as network activations.
arXiv:1912.05015v2 fatcat:h5odp5iyp5fbhe3y22fjif4xyy