OUR People
Academic Staff
Yun Sing Koh
Bio: Yun Sing Koh is a Professor at the School of Computer Science, The University of Auckland, New Zealand. Her main research area is Artificial Intelligence (AI) and Machine Learning (ML). Specifically, focusing on several research strands: continual learning and adaptation, transfer learning anomaly detection, and data stream mining. Yun Sing is passionate about using machine learning for social good, and her research has been applied to interdisciplinary applications in environment and health domains. Yun Sing has published 100+ peer-reviewed publications in top conferences and journals, including IJCAI, IEEE ICDE, IEEE ICDM, Machine Learning Journal and Journal of Artificial Intelligence. She won the New Zealand Royal Society Fast-Start Marsden funding (2018) and the United States Office of Naval Research Grant (2019). Yun Sing has been active in the research community, including serving as the General Co-Chair at the IEEE International Conference on Data Mining 2021, Workshop Co-Chair at ECML/PKDD conference 2021, Program Co-Chair of the Australasian Data Mining Conference 2018 and as the Workshop Co-Chair for the 15th Pacific-Asia Conference on Knowledge Discovery and Data Mining.
Gillian Dobbie
Bio: At school, I excelled in Sciences and discovered my love of Computer Science. I went on to study Applied Engineering at undergraduate and postgraduate level, with a focus on computing and process control. After my undergraduate degree I spent a couple of years in industry working for PEC, programming in assembly language. My PhD research addressed theoretical aspects of database systems. I was one of the first women to complete a PhD in Computer Science at the University of Melbourne. My current research focuses on machine learning, in particular data stream mining and adversarial attacks. My research group creates algorithms that can be used in various application areas, such as predicting peaks and troughs in COVID-19 cases, predicting dementia using routinely collected data, monitoring critical and/or remote sensors, and detecting and defending against various adversarial attacks.
PhD Students
Callum Cory
Bio: Callum is a Ph.D. student working on applying Graph-based Deep Learning models toward Neuroimaging data to help with the identification and understanding of degenerative brain diseases such as Parkinson’s and Alzheimer’s. Challenges in this area include how to effectively downsize a graph into gradually smaller representations, as well as how to combine information from multiple MRI variants, which cover both functional and structural information, into a single model.
Supervisors: Yun Sing Koh, Miao Qiao, Diana-Bernavidas-Prado, Kelly Ke
Di Zhao
Bio: My research focuses on Transfer Learning and Multimodal Reasoning, with an emphasis on applying AI to real-world, interdisciplinary problems. I am particularly passionate about advancing biodiversity monitoring (e.g., animal detection and re-identification) and healthcare applications (e.g., medical question answering) through machine learning. I have a solid publication record in top-tier AI conferences and am deeply committed to teaching and mentorship, fostering and critical thinking in students. I have served as a reviewer for conferences such as ICLR, AAAI, IJCAI.
Supervisors: Yun Sing Koh, Gill Dobbie, Philippe Fournier-Viger
Bowen Chen
Bio: Bowen is a Ph.D. candidate at the University of Auckland. His work focused on measuring the complexity and classification difficulty of instances within a dataset. He recently worked with the National Institute of Weather and Atmosphere (NIWA) to apply transfer learning methods to infer air pollution between different towns in NZ. Currently, his research focuses on applying self-supervised learning to climate modeling.
Supervisors: Yun Sing Koh, Gill Dobbie
Jack Julian
Bio: My work is on continual learning with replay using confidence to inform the most effective time to replay during training.
Supervisors: Yun Sing Koh
Tingrui Qiao
Bio: My research focuses on the analysis and reasoning of multimodal data, particularly visual and textual information. My work has been applied to analyzing user-generated data from the Our Voices program and longitudinal survey data from Growing Up in New Zealand to support the understanding of complex social phenomena, such as school attendance and youth wellbeing. I have served as a reviewer for conferences such as WWW, IJCAI, and ECML.
Supervisors: Yun Sing Koh, Chris Cunningham, Caroline Walker
Yiwen Wang
Bio: My research focuses on continual learning, which enables models to incrementally acquire and retain knowledge across sequential tasks. However, continual learning is fundamentally challenged by catastrophic forgetting, where learning new information interferes with previously acquired knowledge. Addressing this challenge requires mechanisms that balance stability and plasticity. My work explores continual learning frameworks that dynamically adjust learning strategies, aiming to enhance model adaptability while preserving prior knowledge.
Supervisors: Yun Sing Koh, Diana-Bernavidas-Prado
Yinuo Xue
Bio: My research focuses on enhancing user resilience against phishing by leveraging deep learning algorithms. Phishing attacks continue to evolve in sophistication, exploiting subtle linguistic cues and social-engineering tactics to deceive users. Addressing this challenge demands models that not only detect these nuanced patterns in real time but also adapt to emerging threats without degrading performance on previously learned examples. I aim to build systems that empower users to recognize and withstand phishing attempts while maintaining a seamless user experience.
Supervisors: Yun Sing Koh, Giovanni Russello
Yihao Wu
Bio: Yihao is a Ph.D. candidate at the University of Auckland. His research concentrates on Transfer Learning, especially in Domain Generalization (DG), leveraging the capabilities of Large Multimodal Models (LMMs). DG is dedicated to enhancing the generalizability of machine learning models to data from unseen domains by addressing the challenges posed by distribution shifts. He is deeply committed to applying his research to practical, real-world problems, with a special emphasis on wildlife monitoring. His work includes developing wildlife re-identification models robust to distribution shifts to support biodiversity conservation and sustainable development initiatives. Outside school, he has collaborated with prominent institutions such as the Department of Conservation, Manaaki Whenua – Landcare Research, and charity organizations, contributing directly to research tailored to New Zealand’s unique ecological context. He has also served as a reviewer for top-tier conferences, including IJCAI, ACM MM, and ACM WWW.
Supervisors: Yun Sing Koh, Daniel Wilson
Jason Su
Bio: My research focuses on generative models, which learn to approximate complex data distributions and synthesize realistic samples. Diffusion models have emerged as the new state-of-the-art family of deep generative models. They have broken the long-time dominance of generative adversarial networks in the challenging task of image synthesis and have also shown potential in various domains. Currently working on diffusion models for super-resolution.
Supervisors: Yun Sing Koh, Talia Xu
Alumni
- Cristian Gonzalez-Prieto (2025) – Dementia Identification: Machine Learning and Routinely Collected Health Data
- Wernsen Wong (2024) – Preserving Past Knowledge: Improving Memory and Adaptation of Continual Learning Models
- Olivier Graffeuille (2024) – Deep Learning for Data-scarce Environments
- Aaron Keesing (2023) – Emotion Recognition from Speech
- Ocean Wu (2023) – Transfer Learning in Predictive Systems for Data Streams
- Shuxiang Zhang (2023) – Semi-Supervised Concept Drift Detection
- Hongsheng Hu (2022) – On Identifying and Mitigating Against Vulnerabilities of Machine Learning Models
- Ben Halstead (2022) – Adaptive Predictive System for Life Long Learning
- Alex Peng (2021) – Improving the Generalisation of Neural Networks with Unlabelled Data
- Ian Wong (2021) – Efficient Relational Feature Engineering
- Robert Anderson (2020) – Using Meta-Learning to Improve Classification in Data Streams
- Diana Benavides Prado (2019) – A Framework for Long-Term Learning Systems
- Monica Bian (2019) – Heterogeneous Network Mining and Analysis
- David T.J. Huang (2016) – Change Mining and Analysis for Data Streams
- Sidney Tsang (2015) – Fraud Detection in Online Auctions
- Mohammad Abdullatif (2015) – Towards Unsupervised Citation Classification