Program

Monday, October 8, 2018

Time	Event
9:30 – 10:10	Jianqing Fan (Princeton) Gradient Descent with Random Initialization Video Presentation
10:15 – 10:55	Le Song (Georgia Tech) Diverse Neural Network Learns True Target Functions Presentation Slides Video Presentation
11:00-11:30	Coffee Break
11:30-12:10	George Lan (Georgia Tech) Accelerated Stochastic Algorithms for Nonconvex Finite-sum and Multi-block Optimization Video Presentation
12:15-2:00	Lunch and Poster Sessions
2:00 – 2:40	Ding-Xuan Zhou (City U. of Hong Kong) Approximation Theory of Deep Convolutional Neural Networks Presentation Slides
2:45 – 3:25	Ohad Shamir (Weizmann Institute of Science) Optimization Landscape of Neural Networks: Where Do the Local Minima Hide?
3:30 – 4:00	Coffee Break
4:00 – 4:40	Jason Lee (USC) Towards Theoretical Understanding of Overparametrization in Deep Learning Presentation Slides Video Presentation
4:45 – 5:25	Jason Klusowski (Rutgers) Accuracy of High-Dimensional Deep Learning Networks Video Presentation
6:00	Dinner

 

Tuesday, October 9, 2018

Time	Event
9:30 – 10:10	Johannes Schmidt-Hieber (Leiden University) Statistical theory for deep ReLU networks Presentation Slides Video Presentation
10:15 – 10:55	Andrea Montanari (Stanford) A Mean Field View of the Landscape of Two-Layer Neural Networks Video Presentation
11:00-11:30	Coffee Break
11:30-12:10	Rebecca Willett (University of Chicago) Neumann Networks for Inverse Problems in Imaging
12:15-2:00	Lunch and Poster Sessions
2:00 – 2:40	Regina Liu (Rutgers) Prediction with Confidence – a General Framework for Predictive Inference and Uncertainty Quantification
2:45 – 3:25	Tengyu Ma (Stanford) On the Generalization and Approximability in Generative Adversarial Networks (GANs) Presentation Slides
3:30 – 4:00	Coffee Break
4:00 – 4:40	Amitabh Basu (John Hopkins) Bounds on expressibility and training of neural networks
4:45 – 5:25	Po-Ling Loh (Wisconsin) Two results on generalization error
6:00	Dinner

 

Wednesday, October 10, 2018

Time	Event
9:30 – 10:10	Hanie Sedghi (Google) The Singular Values of Convolutional Layers Video Presentation
10:15 – 10:55	Adam Klivans (UT Austin) Efficient Algorithms for Learning One Convolutional Layer Video Presentation
11:00-11:30	Coffee Break
11:30-12:10	Lawrence Carin (Duke University) On Adversarial Learning Video Presentation
12:15-1:15	Lunch
1:20-2:00	Aleksander Madry (MIT) Machine Learning and Security: The Good, the Bad, and the Hopeful Video Presentation
2:00 – 2:40	Animashree Anandkumar (CalTech) SignSGD: A Success Story in Bridging the Gap between Theory and Practice of Deep Learning Presentation Slides
2:45 – 3:25	Deanna Needell (UCLA) Simple classification schemes with binary data Video Presentation
3:30 – 4:00	Coffee Break
4:00 – 4:40	Sebastien Bubeck (MicroSoft Research) Adversarial examples from computational constraints
4:45 – 5:25	Santosh S Vempala (Georgia Tech) On the Complexity of Training a Neural Network Presentation Slides Video Presentation