The Finnish Association for Artificial Intelligence has selected the doctoral dissertation
Theory and Algorithms for Usable Synthetic Data and Efficient Privacy-Preserving Computation
by
Ossi Räisä
as the recipient of the Finnish AI Dissertation Award 2025, awarded annually to an outstanding dissertation in artificial intelligence defended at a Finnish university. The thesis was completed at the Department of Computer Science, University of Helsinki.
Ossi Räisä’s dissertation addresses the fundamental challenge of analysing sensitive personal data in a way that is both statistically rigorous and formally private. Building on differential privacy, the leading mathematical framework for privacy-preserving data analysis, Räisä develops new theory and algorithms for synthetic data generation and efficient privacy-preserving computation. His work combines exceptional mathematical depth with a strong practical orientation, with relevance extending from machine learning to statistics, privacy engineering, and application domains such as health data analysis.
The dissertation is distinguished by the originality and rigor of its contributions as well as an exceptional publication record. Räisä is the first author of all five publications included in the thesis, with work appearing in leading machine learning venues including JMLR, NeurIPS, ICML, and AISTATS. The external evaluations are exceptionally strong and highlight the originality, theoretical depth, and practical significance of the research, with several contributions characterized as groundbreaking. Taken together, the dissertation represents an outstanding scientific achievement with impact extending well beyond a single discipline.