Title: Approximation of functions by neural networks and rational functions

Speaker: Dr Vinesha Peiris
(Curtin Centre for Optimisation and Decision Science)

Abstract:
Function approximation is a fundamental task in numerical analysis and machine learning. Rational functions, formed as ratios of polynomials, have long been valued for their efficiency and accuracy in capturing complex function behaviours. More recently, neural networks – particularly deep learning architectures, have gained prominence due to their versatility and success across diverse application areas.
In this talk, we explore some of these powerful approaches for approximating functions: rational approximation techniques, neural networks and their combinations.

Biography:
Vinesha obtained her honours degree in Mathematics from the University of Sri Jayewardenepura, Sri Lanka in 2018 and later received a PhD from Swinburne University of Technology, Australia in 2022. She formerly served as an associate research fellow at Deakin University before beginning her role as a research fellow at Curtin Centre for Optimisation and Decision Science at Curtin University. In addition to working on optimisation-related large scale industrial projects, she is interested in optimisation, approximation theory, and its uses in neural networks and deep learning. Vinesha has also received several awards from the Australian Mathematical Society including the Bernhard Neumann prize and Maryam Mirzakhani award.

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

WED 30 JULY 2025 15:00-16:00 (AEST, Melbourne Time) **PLEASE NOTE START TIME FOR THIS SEMINAR