ARC@ORU Emerging Malware Threats in the Deep Learning Ecosystem

Venue: Visual Lab, ARC, Örebro University

Remote participation: Join by this Zoom link

No registration is needed.

About the Seminar

Training high-quality deep learning models is a challenging task due to computational and technical requirements. A growing number of individuals, institutions, and companies increasingly rely on pre-trained, third-party models made available in public repositories. These models are often used directly or integrated in product pipelines with no particular precautions, since they are effectively just data in tensor form and considered safe.

This presentation raises awareness of a new machine learning supply chain threat targeting neural networks. We introduce MaleficNet 2.0, a novel technique to embed self-extracting, selfexecuting malware in neural networks. MaleficNet 2.0 uses spread-spectrum channel coding combined with error correction techniques to inject malicious payloads in the parameters of deep neural networks. MaleficNet 2.0 injection technique is stealthy, does not degrade the performance of the model, and is robust against removal techniques. We design our approach to work both in traditional and distributed learning settings such as Federated Learning.

Our aim with this work is to raise awareness against these new, dangerous attacks both in the research community and industry, and we hope to encourage further research in mitigation techniques against such threats.

Speaker

Luigi V. Mancini is full professor at the Sapienza University of Rome, Italy. He is the founder of several Master degree programs in Information and Network Security in the last 20 years. He supervised the doctoral theses on cybersecurity topics of dozens of students in the last 25 years, some of whom are now full professors.

From 2009 to 2012, he served as Head of the PhD Program in Computer Science at the University of Rome “La Sapienza,” overseeing more than 50 doctoral students. From 2013 to 2016, he was Deputy Dean of the Faculty of Information Engineering, Informatics, and Statistics at the same university. From 2017 to 2023, he served as Chair of the Master’s Degree Program (Laurea Magistrale) in Cybersecurity. Since 2015, he has been a Visiting Professor during the summer at the Center for Secure Information Systems, George Mason University, Fairfax, VA (USA).

Luigi V. Mancini has authored more than 150 scientific papers in international conferences and journals. Well-known is his work on "Scalable and efficient provable data possession", which over 15 years ago proposed efficient techniques to ensure data integrity in cloud applications without having to blindly trust the cloud provider. He has served on the program committees of several international conferences. He will be the General Chair of the 31st European Symposium on Research in Computer Security in 2026.The current research interests include: machine learning security, network and information security, and user privacy.

Luigi V. Mancini received the PhD degree in Computer Science from the University of Newcastle, UK, in 1989.