First Conference of Mathematics of AI

The First Conference of Mathematics of AI (MathAI 2025) is a new and timely venue dedicated to the rigorous mathematical foundations of artificial intelligence, addressing the growing need for principled analysis of modern AI systems whose behaviors are often empirically observed but theoretically poorly understood. Topics span optimization theory for deep learning, statistical learning theory, information theory and generalization bounds, the mathematics of neural network expressivity and convergence, and formal analysis of generative models and large language models.

MathAI attracts mathematicians, theoretical computer scientists, and machine learning theorists who want to move beyond empirical observation toward a genuine mathematical science of AI. As AI systems are deployed in high-stakes domains, the demand for provable guarantees about their behavior has intensified, making a dedicated mathematical AI conference not merely academically interesting but practically urgent for the field's long-term reliability and trustworthiness.