Timezone: »

Analyzing Diffusion as Serial Reproduction
Raja Marjieh · Ilia Sucholutsky · Thomas Langlois · Nori Jacoby · Thomas Griffiths

Tue Jul 25 05:00 PM -- 06:30 PM (PDT) @ Exhibit Hall 1 #528

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.

Author Information

Raja Marjieh (Princeton University)
Ilia Sucholutsky (Princeton University)
Thomas Langlois (Princeton University)
Nori Jacoby (Max-Planck Institute)
Thomas Griffiths (Princeton University)

More from the Same Authors