10:00 to 10:30 | Registration | |||
10:30 to 10:40 | Welcome and Introduction |
Jorge Gonzalez-Cazares (National Autonomous University of Mexico (UNAM), IIMAS) Sofia Sanz Del Pino (Isaac Newton Institute) |
||
10:40 to 11:20 | Implicit Compressibility of Overparametrized Neural Networks Trained with Heavy-Tailed SGD | Umut Simsekli (INRIA), (ENS - Paris) | ||
11:20 to 11:35 | Refreshment break | |||
11:35 to 12:15 | Generative Modelling Under Differential Privacy | Sahra Ghalebikesabi (Google) | ||
12:15 to 13:00 | Lunch | |||
13:00 to 13:35 | Moderated discussion: Heavy tails and clipping: benefits and difficulties. | Hao Ni (University College London), (The Alan Turing Institute) | ||
13:35 to 14:15 | Privacy with Synthetic Data: Applications and Developments | Shinpei Nakamura (Yale University) | ||
14:15 to 14:30 | Refreshment break | |||
14:30 to 15:10 | Privacy Properties of Discrete Diffusion Models | Vamsi Potluru (JP Morgan) | ||
15:10 to 15:45 | Moderated discussion: Differential privacy: strategies, benefits and limitations | Bikramjit Das (Singapore University of Technology and Design) | ||
15:45 to 16:00 | Final Remarks and Close | Jorge Gonzalez-Cazares (National Autonomous University of Mexico (UNAM), IIMAS) |