Continual Learning, Continuing the Journey

On 11 July 2025, Do Dinh Hieu successfully defended his thesis, titled "Modeling Multiple Tasks in Recommendation Systems."

The main thesis of this dissertation is that recommender systems must move beyond static offline learning to better accommodate the dynamic and evolving nature of real-world user interactions, preferences, and data environments.

Hieu presented his dissertation in a hybrid mode, with one of the committee members participating remotely

To this end, the dissertation develops three complementary approaches: cross-domain recommendation to facilitate knowledge transfer across datasets, continual learning to adapt to new users, items, and interactions while retaining prior knowledge, and session-aware recommendation to integrate short-term and long-term user preferences.

Hieu joined Preferred.AI in 2020 as a PhD candidate, after completing his undergraduate degree at Vietnam National University, Hanoi.

Dinner at Ellenborough Market Cafe in December 2020 while Hieu was a first-year PhD candidate (Hieu was on the right foreground)

His dissertation research is solidly on recommender systems, appearing in such venues as the ACM Conference on Recommender Systems (RecSys) as well ACM Transactions on Recommender Systems.

Beyond his dissertation research, Hieu is a firm believer in the importance of reproducibility in recommender systems research. This commitment is reflected in his contributions to Cornac, our open-source recommendation framework, as well as his thoughtful reflections in a blog post titled Reproducibility in Recommender Systems: Trust, but Verify. For Hieu, advancing the field is not just about developing better models, but also about ensuring that research findings can be reliably reproduced, rigorously evaluated, and built upon by others.

Post-defense celebration at Clove, Swissotel (Hieu was second from right)

Following his graduation, Hieu will continue his journey with Preferred.AI for a while more as a postdoctoral researcher, working on an industry-funded research project.