PhD Dissertation Proposal by DO Dinh Hieu | Modeling Multiple Tasks in Recommendation Systems

ABOUT THE TALK

In the past decade, artificial intelligence (AI) has achieved remarkable progress, particularly in tasks involving vision and language, thanks to advancements in deep learning. Despite these achievements, recommendation systems—a crucial component of modern information ecosystems—continue to face significant challenges. These systems, integral to personalized content delivery in e-commerce, social media, and other platforms, must evolve to address the dynamic and complex nature of real-world environments.

Traditional recommendation system research primarily focuses on offline learning settings, which simplifies model development and evaluation but falls short in capturing the dynamic nature of real-world scenarios where data streams continuously. This dissertation aims to bridge this gap by formulating more dynamic settings for recommendation systems that can handle new data streams efficiently. Additionally, it explores leveraging multiple tasks to mutually enhance performance and avoid interference.

ABOUT THE SPEAKER

Dinh Hieu Do is currently a Ph.D. candidate at Preferred.AI and SMU School of Computing and Information Systems, advised by Prof. Hady W. Lauw. His primary area of research revolves around the formulation and solution of dynamic real-world recommender systems, including session-based, cross-domain, multi-task, and continual learning settings for recommendations. As a young Ph.D. candidate in the world of research, his thirst for exploration extends beyond academic pursuits. Although he feels inexperienced in both life and research, he yearns to embrace new places and experiences, not only to enrich his life but also to enhance his expertise as a researcher.