The purpose of the International Workshop on Adversarial Machine Learning for Recommendation and Search (AdveRSe) is to provide a venue in the IR community to provide a place to encourage discussion, create novel collaborations, and share research ideas mainly focused on understanding the threat of adversarial machine learning and countering this threat to achieve trustworthy search and recommendation systems. AdveRSe 2021 will take place online on November 1-5, 2021 in conjunction with CIKM 2021.
AdveRSe 2021 offers an opportunity to present and discuss both theoretical and empirical research. Relevant topics include, but are not restricted to:
- Theory and algorithms for AML in search and recommendation
- Attacks on images, videos, audio signals, and text
- Attack.institutions on model parameters with gradient methods
- Attacks by generating of fake profiles, e.g., customer one-commerce platforms
- Attacks crafted with generative adversarial networks
- Real-world attack scenarios–Other attack techniques
- Theory and algorithms for defending with AML in search and recommendation
- Robust optimization methods
- Adversarial training strategies
- Robustness certification
- Generative adversarial networks to protect from adversarial attacks
- Data augmentation with adversarial training
- Datasets for evaluating model robustness
- Evaluation of Adversarial Attacks and Defences in search and recommendation
- Offline evaluation measures and protocols
- Online evaluation measures and protocols
- Advanced applications (Trustworthy ML) for search and recommendation
- Federated machine learning
- Explainable machine learning models
- Causal and Counterfactual reasoning
Submissions of full research papers must be in English, in PDF format in the CEUR-WS two-column conference format available at CEURART or at OVERLEAF TEMPLATE if an Overleaf is preferred.
- Long Papers should report on substantial contributions of lasting value. The Long papers must have a length of minimum 6 and maximum 8 pages (plus an unlimited number of pages for references). Each accepted long paper will be included in the CEUR on-line Workshop proceedings and presented in a plenary session as part of the Workshop program.
- Short/Position Papers typically discuss exciting new work that is not yet mature enough for a long paper. In particular, novel but significant proposals will be considered for acceptance to this category despite not having gone through sufficient experimental validation or lacking strong theoretical foundation. Applications of adversarial learning in recommendation and search systems to novel areas are especially welcome. The Short Papers must have a length of minimum 3 and maximum 5 pages (plus an unlimited number of pages for references). Each accepted short paper will be included in the CEUR on-line Workshop proceedings and presented in a plenary session as part of the Workshop program
To be considered, papers must be received by the submission deadline (see Important Dates). The short and long papers review process is double-blind. The authors must anonymize their submissions. Submitted papers will be evaluated according to their originality, technical content, style, clarity, and relevance to the workshop. Short and long paper submissions must be original work and may not be under submission to other venues at the time of review (authors can upload to institutional or other preprint repositories such as arXiv.org before reviewing is complete ).
Submission will be through Easychair at https://easychair.org/conferences/?conf=adverse2021.
- Submission website opens: May 26, 2021
- Submission deadline: July 10, 2021 (Extended to July 17, 2021)
- Notification of acceptance: August 10, 2021
- Camera ready deadline: TBA
- AdvRSe 2021: November 1-5, 2021
Deadlines refer to 23:59 (11:59pm) in the AoE (Anywhere on Earth) time zone.TB