Workshop Program

For each paper presentation, the allocated time is 20 minutes total (15 minutes for the presentation + 5 minutes for Q&A).

Here is the detailed schedule for the session:

14:30 – 14:45 | Opening Remarks Introduction to the DeepFake problem and Workshop Program overview. • Chair: Duccio Meconcelli

14:45 – 15:15 | Keynote FRIDA: Fake image Recognition and source Identification via Diffusion features Analysis • Speaker: Simone Bonechi

15:15 – 15:35 | Paper Presentation (967) AgentForensics: An Open-Protocol Multi-Agent Framework for Synthetic Image Analysis • Authors: Paolo Bestagini, Mohamed Aziz Cherif, Ashkriti Dewan, Daniele Ugo Leonzio, Sara Mandelli, Moro Moro, Davide Redaelli, Vivek Kumar Sharma

15:35 – 15:55 | Paper Presentation (932) On the Robustness of AI-Generated Image Attribution Systems Under Adversarial Attacks • Authors: Nasrin Malekzadeh Goradel, Niccolò Pancino, Yaser Gholizade Atani, Benedetta Tondi, Mauro Barni

16:00 – 16:30 | Coffee Break (30 Min)

16:30 – 17:00 | Challenge Overview & Rankings Presentation of the Challenge, dataset, and final rankings. • Chair: Pietro Bongini

17:00 – 17:20 | Paper Presentation (961) Probing Stable Diffusion 3.5: A Comparative Exploration of Internal Features for Synthetic Image Source Attribution • Authors: Duccio Meconcelli, Paolo Andreini, Sara Bacconi, Simone Bonechi, Pietro Bongini, Franco Scarselli, Monica Bianchini

17:20 – 17:40 | Paper Presentation (933) Angular-Margin Representation Learning for Synthetic Image Source Attribution under Controlled Post-Processing Shifts • Author: Tsuyoshi Tanaka

17:40 – 18:00 | Paper Presentation (949) Restoring the Hidden Degradation Chain: Post-Processing-Aware Training for Robust Synthetic Image Source Attribution • Author: Lev Kokovkin