Generalization Gap in Amortized Inference

05/23/2022
by   Mingtian Zhang, et al.
6

The ability of likelihood-based probabilistic models to generalize to unseen data is central to many machine learning applications such as lossless compression. In this work, we study the generalizations of a popular class of probabilistic models - the Variational Auto-Encoder (VAE). We point out the two generalization gaps that can affect the generalization ability of VAEs and show that the over-fitting phenomenon is usually dominated by the amortized inference network. Based on this observation we propose a new training objective, inspired by the classic wake-sleep algorithm, to improve the generalizations properties of amortized inference. We also demonstrate how it can improve generalization performance in the context of image modeling and lossless compression.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
09/04/2021

On the Out-of-distribution Generalization of Probabilistic Image Modelling

Out-of-distribution (OOD) detection and lossless compression constitute ...
research
11/12/2020

VCE: Variational Convertor-Encoder for One-Shot Generalization

Variational Convertor-Encoder (VCE) converts an image to various styles;...
research
05/28/2018

Block-optimized Variable Bit Rate Neural Image Compression

In this work, we propose an end-to-end block-based auto-encoder system f...
research
06/07/2020

Improving Inference for Neural Image Compression

We consider the problem of lossy image compression with deep latent vari...
research
06/18/2020

A Tutorial on VAEs: From Bayes' Rule to Lossless Compression

The Variational Auto-Encoder (VAE) is a simple, efficient, and popular d...
research
03/05/2019

Improving Generalization of Deep Networks for Inverse Reconstruction of Image Sequences

Deep learning networks have shown state-of-the-art performance in many i...
research
06/30/2021

Exploring Context Modeling Techniques on the Spatiotemporal Crowd Flow Prediction

In the big data and AI era, context is widely exploited as extra informa...

Please sign up or login with your details

Forgot password? Click here to reset