Benign Loss Landscapes Can Coexist with Worst-Case Hardness @FieldsInstitute
Benign Loss Landscapes Can Coexist with Worst-Case Hardness  @FieldsInstitute
Uploaded September 2026 | Updated November 2026, 13 hours ago
Speaker: Zach Furman, University of Melbourne
Date: September 14, 2026
Abstract: fields.utoronto.ca/talks/Benign-Loss-Landscapes-Can-Coexist-Worst-Case-Hardness

2026-2027 Mathematical AI Seminar: fields.utoronto.ca/activities/26-27/mathai

Centre for Mathematical AI: fields.utoronto.ca/centres/centre-mathematical-ai
Benign Loss Landscapes Can Coexist with Worst-Case HardnessLecture 01 | Mathematics for AI SafetyHolder Estimates for Complex Hessian EquationsMaking Progress on Multi-Period Financial Models: Lessons from Reinforcement LearningSimplicial Distributions and Quantum ContextualityLecture 13 | Mathematical Foundations of AILecture 24 | Mathematical Foundations of AIProjections between Ramsey quasi-orders and topological Ramsey spaces (Part 1)Higher special Lindelöf trees from the proxy principleLecture 04 | Mathematical Foundations of AIGradients, Groups, Grids: How neural networks learn features in algebraic tasksGeometric inequalities on manifolds via optimal transport
Fields Institute |

Benign Loss Landscapes Can Coexist with Worst-Case Hardness

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER