Special Sessions

Accepted Special Sessions

We are pleased to announce the accepted special sessions:

  • Forensic Detection of AI Synthesized Media
  • Lessons Learned from the ALASKA2 Image Steganalysis Challenge

Forensic Detection of AI Synthesized Media

Organizers:

  • Siwei Lyu (University of Albany, USA)
  • Luisa Verdoliva (University Federico II of Naples, Italy)

Motivation

The past few years have seen a troubling rise of fabricated or manipulated images, audios, and videos backed by advanced AI technologies, in particular, deep neural networks (DNNs). Notable examples of DeepFakes include DNN-generated face swapping videos and GAN synthesized realistic human face images. While there are interesting and creative applications of this technology, it can be weaponized for serious political, social, financial, and legal consequences. Left unchecked, AI-synthesized media can escalate the scale and danger of disinformation.

This Special Session offers a timely venue for novel research contributions to forensic detection technologies for AI synthesized media. The scope of works in this Special Session will include, but not limited to:

  • Detection and attribution of DeepFake videos and audios;
  • Detection and attribution of GAN generated images;
  • Adversarial models to detection methods of AI synthesized media;
  • New benchmarks and datasets for AI synthesized media detection;
  • Obstruction and disruption of AI media synthesis pipeline.

Lessons Learned from the ALASKA2 Image Steganalysis Challenge

Organizers:

  • Rémi Cogranne (Troyes University of Technology, France)
  • Quentin Giboulot (Troyes University of Technology, France)
  • Patrick Bas (CNRS, École Centrale de Lille, France)

Motivation

Over the past decades, steganography and steganalysis have been developed together in a sort of "cat and mouse game". While the former aims at hiding data into cover media in a more and more "secure" way, it benefits from steganalysis that seeks at designing hidden data detection method with lower and lower classification error probabilities.

Over the plethora of academic works in both steganography and steganalysis, the current works, especially in steganalysis, have fundamental limitations when it comes to practical applications. The BOSS base has been extensively used while it is very singular (with all image harshly downsized, grayscale images from only 7 different camera). Hence, both steganography and steganalysis seem to be tailored to specific dataset, embedding scheme and detection techniques.

The goal of the competition is to push the community towards image steganalysis, "into the wild" facing the following challenges:

1. The very high heterogeneity of image "sources", with widely different type of content, processing operations, sensor noises, JPEG quality factors, etc.

2. A broad variability of data hiding, with three different embedding scheme and a payload that is different for each and every media

3. Designing machine learning method for detection with low false positive, ideally that could be guaranteed.

The goal of this special session is both to allow participants with high scores to present their work and their finding as well as to draw lessons from tat we hope will shape the research in steganalysis.

Call for Special Sessions

The call for special sessions is already closed.

Call for Special Session Proposals, WIFS 2020

We welcome proposals for special sessions at WIFS 2020 to complement the regular technical program. A special session consists of about 5 papers, with a unifying theme that focuses on fields of special and emerging interest to the information forensics and security community.

Proposal Submission: ​Prospective organizers of Special Sessions should submit proposals with the following information:

  • Title of the Special Session
  • A description of the significance and novelty of the topic and the rationale for the proposed session including any possible interdisciplinary flavor
  • Names, affiliations and a short biography of the organizers highlighting their expertise in the topic of the Special Session
  • A list of research groups currently active in that area and potential contributors
  • A list of 5 potential contributed paper submissions (if available including provisional titles, authors, contact information of the corresponding author, and a short abstract).

Proposals and inquiries should be sent via e-mail to the Special Sessions chairs (proposals should be in single PDF attachments):

  • Shih-Fu Chang, Columbia University, sc250 [at] columbia [dot] edu
  • Xu Zhang, Amazon, xzhnamz [at] amazon [dot] com

Deadlines:

  • Special Session proposals due: May 10, 2020
  • Notification of acceptance of Special Sessions: May 17, 2020
  • Paper submission to accepted Special Sessions: June 30, 2020

Evaluation Criteria: Proposals will be evaluated based on the timeliness and relevance of the topic and the potential for obtaining enough high-quality contributions.

Special Session Paper Submission Process: Upon acceptance of the special session proposal, their topics and organizers will be publicized via the WIFS 2020 website. Organizers are expected to invite the potential contributors to submit papers in the same format as regular papers. Organizers should not contribute more than one paper to their own Special Session. Special sessions should be open to new submissions that are relevant to the topics, but not on the initial list of potential contributions in the special session proposal.

Invited Position Paper: ​Organizers of special sessions will be given the opportunity to submit a two-page invited position paper that introduces the special session. The position paper, clearly marked as such, will be included in the proceedings. Quality assurance will be done by the Special Session Chairs.

Paper Review Process: Submissions to accepted Special Session will undergo the same review process as regular submissions. The organizers are invited to suggest reviewers for the submissions to the special session, and additional reviewers will be selected from the Program Committee by the Program Chairs. The Special Sessions are part of the main conference. Hence, the review schedule and the camera-ready paper submission date are the same. Acceptance decisions will be taken by the Program Committee.

It is expected that the organizers of accepted special sessions attend the workshop to chair their session. Further, to ensure the coherence of the program and that papers addressing similar issues are grouped together, at the discretion of the Program Committee, individual papers may be removed from or added to a Special Session. If too many submissions to an accepted Special Session do not pass the review process, and we are unable to find suitable substitutes from the regular papers, the Special Session may be canceled and selected papers from the canceled sessions may be placed into the regular program.