A Principled Self-Referenced Early Stopping Approach for Deep Image Prior

Chaoyan Huang, Cheng-Han Huang, Ismail R. Alkhouri, Rongrong Wang · 2026

Read A Principled Self-Referenced Early Stopping Approach for Deep Image Prior by Chaoyan Huang, Cheng-Han Huang, Ismail R. Alkhouri, Rongrong Wang (2026) 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-2605-25299.txt.

Description

Chaoyan Huang and colleagues tackle a flaw in Deep Image Prior, which rebuilds damaged images with a network trained on the damaged image alone. Because the network is oversized, it memorizes noise, so early stopping matters. The best stopping rule watches wobbles in output variance, but those wobbles can arrive too soon. The team shows good stopping is possible with two independent noisy copies, then builds stand-ins from the image itself. Their three methods beat stopping rules in restoration and medical tasks, with no noise estimate needed.

Catalog record

CREATORChaoyan Huang, Cheng-Han Huang, Ismail R. Alkhouri, Rongrong Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE98647
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.25299

Browse the full library · About the Tower of Babel