DVAR: Adversarial Multi-Agent Debate for Video Authenticity Detection DOI: XXXXXXX.XXXXXXX Conference: ; ; CCS: Computing methodologies Scene anomaly detection

Hongyuan Qi, Feifei Shao, Ming Li, Hehe Fan, Jun Xiao · 2026

Read DVAR: Adversarial Multi-Agent Debate for Video Authenticity Detection DOI: XXXXXXX.XXXXXXX Conference: ; ; CCS: Computing methodologies Scene anomaly detection by Hongyuan Qi, Feifei Shao, Ming Li, Hehe Fan, Jun Xiao (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-2604-16987.txt.

Description

Qi and four colleagues turn video authenticity detection into an open debate between two agents: one arguing the footage was generated, the other defending a natural explanation. Through rounds of cross-examination, each side must answer abnormal evidence, and Occam's Razor, via minimum description length, judges which account carries the smaller logical burden. The training-free DVAR system matches supervised leaders while holding up better to unseen generators, and leaves a clear reasoning trail.

Catalog record

CREATORHongyuan Qi, Feifei Shao, Ming Li, Hehe Fan, Jun Xiao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42259
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.16987

Browse the full library · About the Tower of Babel