Published:
Author: Salena Fitzgerald

Research from the Department of Applied Mathematics and Statistics will be presented at the 2022 NeurIPS (Neural Information Processing Systems) conference set for Thursday, Nov. 29 through Friday, Dec. 9 in New Orleans. 

An annual machine learning and computational neuroscience conference, NeurIPS is considered a flagship event for the machine learning and AI communities. Only 25.7 % of papers submitted to this conference were accepted by the NeurIPS program committee. Researchers will present a poster in person, as well as a recorded video presentation that will be included in the virtual program. 

 

AMS research to be presented includes: 

Dynamics of SGD with Stochastic Polyak Stepsizes: Truly Adaptive Variants and Convergence to Exact Solution 

Nicolas Loizou, assistant professor, with Antonio Orvieto and Simon Lacoste-Julien 

 

Convergence for score-based generative modeling with polynomial complexity 

Holden Lee, assistant professor, with Jianfeng Lu and Yixin Tan 

 

Global Linear and Local Superlinear Convergence of IRLS for Non-Smooth Robust Regression 

Christian Kümmerle, former postdoctoral fellow, with Liangzu Peng and René Vidal, Herschel L. Seder Professor.

 

Learning to Compare Nodes in Branch and Bound with Graph Neural Networks 

Abdel Ghani Labassi, PhD Candidate, with Didier Chetelat and Andrea Lodi