Bias and Generalization in Deep Generative Models: An Empirical Study

TL;DR

This paper systematically investigates the inductive bias and generalization capabilities of deep generative models for images, revealing insights into their ability to generate novel attributes and their similarities to human cognition.

Contribution

It introduces a new framework inspired by cognitive psychology to analyze bias and generalization in deep generative models, providing empirical insights into their behavior.

Findings

Models exhibit biases similar to human psychology.

Patterns of attribute generation are consistent across architectures.

The framework reveals when and how models generate novel features.

Abstract

In high dimensional settings, density estimation algorithms rely crucially on their inductive bias. Despite recent empirical success, the inductive bias of deep generative models is not well understood. In this paper we propose a framework to systematically investigate bias and generalization in deep generative models of images. Inspired by experimental methods from cognitive psychology, we probe each learning algorithm with carefully designed training datasets to characterize when and how existing models generate novel attributes and their combinations. We identify similarities to human psychology and verify that these patterns are consistent across commonly used models and architectures.