Aaron Gibson-John, Liam Heng and Rui Tong
Speaker 1: Aaron Gibson-John
Title: Predictive Quality in Manufacturing, with Application to Thermal Inkjet Printheads
Abstract: Predicting final product quality from stage-wise manufacturing data is a supervised learning problem with two complications. The in-line manufacturing processes are nonstationary, and the labels come from acceptance criteria that are revised over time. These challenges are discussed in the context of thermal inkjet printhead manufacturing, along with existing methods for addressing each.
Speaker 2: Liam Heng
Title: Data-driven methods to support national team selection
Abstract: As with other invasion sports, football is difficult to analyse because of its continuity, dependence on other actors, and playing context. Therefore, despite the inundation of data, we find that even modern performance metrics and analytical methods do not capture the complexity of the sport. We aim to develop models to compare players in a context-agnostic way, quantify the quality of interaction, and value teams, with an eye towards a team selection framework. In this talk, we further discuss the challenges facing research and industry and introduce our initial progress.
Speaker 3: Rui Tong
Title: Towards Automating Industrial Print Quality Assessment (PQA) via Statistics and ML
Abstract: In printhead production, the final testing stage generally involves physical operation of the printhead, and inspection of printed output for defects. We aim to develop statistics and machine learning driven PQA procedures for Memjet’s production line. Image quality and colour representation are briefly discussed as proxy for automated PQA. An initial attempt with optimal transport is discussed.
Statistics seminar
School of Mathematics and Statistics, UNSW
Friday, 13 November 2026, 4:00 pm
Microsoft Teams/ Anita B. Lawrence 4082