Title: Faithful-Newton: a variant of Newton’s method for large-scale optimization

Speaker: Alexander Lim
University of Queensland

Summary:
In machine learning (ML), gradient-based algorithms have long been the default choice for solving empirical risk minimization problems. In this talk, I will provide a brief overview of an alternative approach, Newton’s method, discuss why it is not commonly used in ML, and introduce one possible way to make it more practical and effective in modern ML applications.

Bio:
Alexander Lim completed his PhD in Mathematics at the University of Queensland (UQ), under the supervision of Dr Fred Roosta. His research focuses on developing and analyzing second-order type methods for large-scale problems. Prior to his PhD, he obtained a Bachelor of Mathematics from UQ in 2020, and a Bachelor of Arts (Honours), major in philosophy of mathematics and physics, from the University of Tasmania in 2016.

JOIN ON ZOOM – MEETING ID: 873 1557 5255; PASSWORD: 778635

WED 15 OCTOBER 2025 16:00-17:00 (AEDT, Australian Eastern Daylight Savings Time)