Cyber security provider F-Secure will research advancements in trustworthy artificial intelligence (AI) as part of SPATIAL (Security and Privacy Accountable Technology Innovations, Algorithms, and Machine Learning).

The main F-Secure’s objectives will be to design, develop and evaluate methods for enhancing the resilience of AI-based technologies utilized in the cybersecurity domain.

“Secure AI is the foundation for trustworthy AI,” said F-Secure Vice President of Artificial Intelligence Matti Aksela.

Founded in 1988, F-Secure is the global leader in providing security as a service through telecommunication operators and is a major global player in the advanced threat protection and cybersecurity services domains. Our award-winning products protect people and companies against everything from crimeware to corporate cyberattacks, and are available from over 6000 resellers and 200 Internet Service Provider and Mobile Network Operator partners in more than 40 countries. F-Secure’s top-class security consultants deliver incident response, digital forensics, penetration testing, security design, vulnerability assessments of web applications and networks, and PCI DSS services.

Samuel Marchal is the Principal Investigator from F-Secure for SPATIAL

Samuel Marchal is a Senior Data Scientist and a Team Lead specializing in the security of machine learning systems. He was awarded with the IEEE ICDCS 2017 Best Poster/Demo Award in 2017. He has vast experience in System Security and Machine Learning, and has contributed to many scientific peer-reviewed publications, e.g., ” DAWN: dynamic adversarial watermarking of neural networks”, “SAFELearn: Secure Aggregation for private FEderated Learning”, “PRADA: protecting against DNN model stealing attacks” or “AuDI: Toward Autonomous IoT Device-Type Identification Using Periodic Communication”.

His wide experience in security related projects (e.g. “Securing Lifestyle of Internet-of-Things” or “Contextual Security: Balancing Security and Usability via Context Inference”) makes him a valuable investigator for the SPATIAL project.

 

Principal Investigator

  1. How did you started/joined/enrolled SPATIAL?

At F-Secure, we want to ensure that machine learning can be used without trading-off security. SPATIAL offers us to advance the security of machine learning together with industrial and academic partners having the same objective.

  1. What are your expectations in a project of this nature?

I expect the project to advance the understanding of machine learning systems from the security and explainability perspectives. Altogether, SPATIAL will help increase the trustworthiness of machine learning and AI.

  1. What can the research community expect from SPATIAL?

It can expect an industrial perspective on securing and explaining machine learning. Most of the research on machine learning security so far has been focused on ML models used in tasks around image analysis and classification, which often does not generalize to other applications. In SPATIAL, we will study machine learning systems actually deployed and applied in such important domains as IoT, cybersecurity and 5G.

  1. Where do you see SPATIAL results in 10 years?

I see the outcomes of SPATIAL being used in the industry as guidelines and technical solutions to implement secure and explainable machine learning systems.