Conference Description
Key Takeaways
- Academic conference addressing vulnerabilities, robustness, and trust in machine learning systems
- Covers attacks and defenses, privacy, fairness, interpretability, and ML system verification
- Designed for researchers, PhD students, and industry professionals in ML, AI, and cybersecurity
- Features research presentations, workshops, and competitions
- Sponsored by the IEEE Computer Society Technical Committee of Security and Privacy
About the Conference
The 5th IEEE Conference on Secure and Trustworthy Machine Learning (SaTML 2027) brings together academic researchers and industry practitioners to address the security challenges facing machine learning systems. Now in its fifth year, the conference provides a platform for presenting new research findings, systematizing existing knowledge, and debating emerging challenges in ML security. The event takes place in Reykjavik, Iceland, and includes a main conference programme alongside workshops and competitions.
Technical Focus Areas
SaTML 2027 examines both theoretical foundations and practical implementations of secure machine learning. The programme addresses how ML systems can be attacked and defended, covering adversarial techniques and countermeasures. Verification of ML algorithms and systems forms a significant component, alongside forensic analysis methods for investigating ML behaviour.
Privacy considerations in machine learning receive dedicated attention, examining how models can leak sensitive information and techniques for preserving confidentiality. The conference also explores fairness and interpretability, two areas increasingly relevant as ML systems are deployed in consequential decision making. Trustworthy data curation addresses the foundational challenge of ensuring training data integrity.
Trustworthy ML in Cybersecurity Applications
A dedicated track examines the application of trustworthy machine learning within cybersecurity contexts. This intersection is particularly relevant as ML systems are increasingly deployed for threat detection, malware analysis, and security automation. The conference addresses how security teams can build confidence in ML tools while understanding their limitations and potential failure modes.
Who Should Attend
The conference serves academic researchers and PhD students working in machine learning, artificial intelligence, security, and privacy. Industry professionals including security analysts, data scientists, ML engineers, and technical leads will find relevant content on practical implementation challenges. The programme accommodates participants ranging from graduate students to senior researchers and technical executives. Organizations in technology, cybersecurity, and research institutions are represented across the attendee base.
Event Format
SaTML 2027 is structured as an in-person event with research presentations forming the core programme. Workshops provide focused exploration of specific topics, while competitions offer hands-on engagement with ML security challenges. The format emphasizes technical depth and interactive participation rather than commercial exhibition.