Research

We study the mathematical foundations and practical deployment of trustworthy AI, with a focus on federated learning (FL), explainability, robustness, and governance.

Mathematical Theory of Trustworthy AI

Funding: Research Council of Finland (RCF) & JAES

This project develops rigorous principles for trustworthy federated learning that go beyond accuracy to incorporate fairness, interpretability, robustness, and accountability. We build theory and methods for human-centric FL systems that protect agency and fundamental rights while enabling personalized models.

  • Foundations for trustworthy FL (regularization, constraints, and risk trade-offs).
  • Personalized explainability metrics aligned with user notions of similarity.
  • Design criteria that connect mathematical guarantees to real-world deployment.

FLAIG — Forward-Looking AI Governance in Banking & Insurance

Funding: Business Finland (Co-research). Partners: University of Turku, Aalto University, Tietoevry, OP Financial Group, Veritas.

FLAIG develops sector-specific AI governance frameworks, auditing principles, and personalized explainability solutions for high-risk finance applications in the era of generative AI. The project advances compliance with the EU AI Act and operationalizes fairness, transparency, and accountability for real-world decision systems.


For a complete list of publications, see our Google Scholar profile ↗.