Santiago AranguriI am a third year mathematics PhD student at New York University (Courant Institute), working with Eric Vanden-Eijnden on phase transitions and scaling limits of diffusion models and Arthur Jacot on scaling laws for neural networks. I obtained my B.S. in Mathematics at Stanford University in 2022, where I worked on interacting particle systems with Amir Dembo for my undergraduate thesis. My work focuses on making progress on machine learning by understanding some of its fundamental mechanisms using tools from high-dimensional probability and statistical mechanics. Here is my CV You can contact me at aranguri [at] nyu [dot] edu PublicationsOptimizing Noise Schedules of Generative Models in High DimensionsS. Aranguri, G. Biroli, M. Mezard, E. Vanden-Eijnden arXiv preprint arXiv:2501.00988. Under review. Phase-aware Training Schedule Simplifies Learning in Flow-Based Generative Models S. Aranguri, F. Insulla arXiv preprint arXiv:2412.07972. Under review. Mixed Dynamics In Linear Networks: Unifying the Lazy and Active Regimes Z. Tu, S. Aranguri, A. Jacot Advances in Neural Information Processing Systems, 2024. Untangling planar graphs and curves by staying positive S. Aranguri, H. Chang, D. Fridman Proceedings of the 2022 Annual ACM-SIAM Symposium on Discrete Algorithms, 211–225. |