Scalable and Lightweight Content-based Recommender Systems

The School of EECS is hosting the following PhD Progress Review 3 Seminar:

Scalable and Lightweight Content-based Recommender Systems

Speaker: Hung Vinh Tran

Host: A/Prof. Sen Wang

Abstract:

Recommender systems have become a fundamental component of modern online platforms for users to navigate enormous volumes of content and products. Modern content-based recommender systems enhance their predictive accuracy by incorporating side features such as item textual features and user information, but this introduces substantial overhead throughout the model lifecycle involving two key stages: training and deployment. First, recommender systems must learn from continuously changing user interactions while serving models on a variety of deployment devices. Repeated model training leads to revisiting large interaction datasets, especially in multimodal settings that require a deep multimodal encoder. Then, at deployment, a heavily parameterized recommendation model must fit devices with different memory budgets. This thesis addresses these limitations through two connected objectives: data efficiency during repeated training and parameter efficiency at deployment.

In summary, the new solutions developed in this thesis enable recommender systems to efficiently adapt to both changing user behavior and diverse deployment constraints, reducing the cost of frequent retraining while supporting flexible operation under different memory budgets. More broadly, they make advanced content-based recommendation more accessible in resource-constrained settings, where limited computation, memory, or training infrastructure would otherwise make frequent retraining and deployment impractical. Together, these contributions provide a scalable path toward content-based recommender systems that are cheaper to build, store, and maintain without sacrificing usability. 

Bio: Hung Vinh Tran (Trần Vinh Hưng) is a final-year PhD student in Computer Science at the University of Queensland, Australia. He is currently working on model efficiency for recommender systems, under the supervision of Assoc. Prof. Rocky Tong Chen, Prof. Hongzhi Yin, and Dr. Xia Xin. Previously, he was a researcher and engineer at Cinnamon AI. He obtained his Bachelor's Degree from the Honors Program of the Faculty of Information Technology, Ho Chi Minh City University of Science, Vietnam National University (VNU-HCMUS).