Analyzing Diffusion as Serial Reproduction

TL;DR

This paper establishes a theoretical link between diffusion models in machine learning and serial reproduction in cognitive science, explaining their properties and sensitivities through this analogy, supported by simulations.

Contribution

It introduces a novel theoretical framework connecting diffusion models to serial reproduction, providing insights into their behavior and properties.

Findings

Diffusion models are analogous to serial reproduction in cognitive science.

The properties of diffusion models can be explained through this correspondence.

Simulations support the theoretical insights.

Abstract

Diffusion models are a class of generative models that learn to synthesize samples by inverting a diffusion process that gradually maps data into noise. While these models have enjoyed great success recently, a full theoretical understanding of their observed properties is still lacking, in particular, their weak sensitivity to the choice of noise family and the role of adequate scheduling of noise levels for good synthesis. By identifying a correspondence between diffusion models and a well-known paradigm in cognitive science known as serial reproduction, whereby human agents iteratively observe and reproduce stimuli from memory, we show how the aforementioned properties of diffusion models can be explained as a natural consequence of this correspondence. We then complement our theoretical analysis with simulations that exhibit these key features. Our work highlights how classic paradigms in cognitive science can shed light on state-of-the-art machine learning problems.