The paper presents a predictive theory of creativity in convolutional diffusion models, identifying inductive biases that enable novel image generation beyond training data through local patch combinations.

https://arxiv.org/abs//2412.20292

YouTube: https://www.youtube.com/@ArxivPapers

TikTok: https://www.tiktok.com/@arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers