The Department at AISTATS 2026

We are delighted to share that two of our faculty members, Zoltán Szabó and Milan Vojnovic, have had papers accepted to this year’s AISTATS conference.

Professor Szabó’s paper 'The Minimax Lower Bound of Kernel Stein Discrepancy Estimation', with collaborators José Criberio-Ramallo, Agnideep Aich, Florian Kalinke, and Ashit Aich, explores a fundamental question in modern machine learning: how well we can measure differences between complex probability distributions. Their paper establishes theoretical limits on the accuracy of a widely used method known as Kernel Stein Discrepancy. In basic terms, their work helps researchers understand how reliable these tools are when comparing models, an essential task in areas like generative AI and statistical testing.

Meanwhile, Professor Vojnovic and his co-authors Junghyun Lee, Kyoungseok Jang, Kwang-Sung Jun, and Se-Young Yun, will present a new method for analyzing large, complex datasets where the underlying structure is ‘low-rank’, a common scenario in recommendation systems and network data. Their approach, ‘GL-LowPopArt’, outlined in 'GL-LowPopArt: A Nearly Instance-Wise Minimax-Optimal Estimator for Generalized Low-Rank Trace Regression' is designed to produce highly accurate estimates tailored to individual data instances, pushing closer to the theoretical limits of performance.

Together, these contributions highlight the strength of our faculty’s research and their active role in advancing the foundations of modern data science.

The 2026 edition of AISTATS marks the 29th installment of the conference and will take place in Tangier, Morocco, bringing together an international community of academics and practitioners between 2-5 May.