Spectral Analysis of Diffusion Models with Application to Schedule Design
Roi Benita, Michael Elad, Joseph Keshet · 2025
Read Spectral Analysis of Diffusion Models with Application to Schedule Design by Roi Benita, Michael Elad, Joseph Keshet (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-00180.txt.
Description
Roi Benita, Michael Elad, and Joseph Keshet analyze diffusion models in 2025 through the lens of frequency. These models start from pure noise and clean it step by step into images or sound. Some design choices, like how fast noise fades at each step, still rest on habit rather than theory.
Assuming the noise behaves in a standard way, the team derives a closed-form description of how each frequency in the signal evolves during generation. The formula turns schedule design into a matching problem: shape the noise path to the data's own frequency mix. The view explains why some schedules practitioners favor actually work.
Catalog record
| CREATOR | Roi Benita, Michael Elad, Joseph Keshet |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 129 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-00180