All talks listed in Pacific Time. Schedule subject to change.

–
Coffee and Check-In
–
Opening Remarks
–
Detection and Steering in LLMs using Feature Learning
Daniel Beaglehole (UC San Diego)
–
Break
–
Adam and friends via duality
Matus Telgarsky (NYU)
–
Lunch (on your own)
–
Theory of Scaling: A General Framework for Scale-aware Training
Soufiane Hayou (UC Berkeley)
–
Break
–
A Theory of Learning with Autoregressive Chain of Thought
Nirmit Joshi (TTIC)
–
Reception and Poster Session (2nd Floor)
–
Coffee and Check-In
–
A Statistical Theory of Contrastive Pre-training and Multimodal Generative AI
Song Mei (UC Berkeley)
–
Break
–
Generalization and overfitting in two-layer neural networks
Pierfrancesco Urbani (CNRS)
–
Lunch (on your own)
–
Emergence and scaling laws for SGD learning and Learning Compositional Functions with Transformers
Jason Lee (Princeton University)
–
Break
–
Local minima of the empirical risk in high dimension
Basil Saeed (Stanford University)
–
Weak-to-Strong Generalization Even in Random Feature Networks, Provably
Marko Medvedev (University of Chicago)
–
Happy Hour at Freehouse
–
Coffee and Check-In
–
Benefits of Early Stopping in Gradient Descent for Overparameterized Logistic Regression
Jingfeng Wu (UC Berkeley)
–
Break
–
Neural networks for PDE solvers: representational power and inductive biases
Andrej Riteski (Carnegie Mellon University)
–
Lunch (on your own)
–
Pencil: Long Thoughts with Short Memory
Zhiyuan Li (TTIC)
–
Break
–
Positive Distribution Shift
Nati Srerbo (TTIC)
–
Panel Discussion
Bin Yu (UC Berkeley), Andrej Riteski (Carnegie Mellon University), Matus Telgarsky (NYU)