Memorable Moments
Top Rankings at TREC iKAT 2025 Competition
Thrilled to share that our team achieved excellent results in the TREC iKAT 2025 competition! Full Results Official Overview
Together with my teammates Yuan-Ching Kuo, Tsung-En Yu, Yu-Cheng Chang, Guan-Wei Yeo, Quah Eugene, and Fan-Jie Shih from Dr. Chuan-Ju Wang's laboratory, we achieved strong results across multiple run-wise evaluations:
- Offline Passage Ranking: 5th place
- Offline Generated Response 1 (RAG): 13th place
- Offline Generated Response 2 (Generation-only): 1st place
- Online Interactive Response Generation: 4th place
Presenting our FincGAN paper at ICASSP 2024
I was delighted to participate in ICASSP 2024. I met many exceptional researchers there, including experienced professors and enthusiastic PhD students, who provided valuable guidance and inspired new research ideas.
During the week-long conference, I attended numerous lecture sessions. One session particularly stood out—after the presenter finished, no other participants asked questions or gave feedback. I ended up being the only one to ask a question, which meant I had to listen extremely carefully to avoid asking something stupid. Overall, it was a memorable experience and I learned a great deal from the conference.
My Master's Graduation Ceremony – Top of the Class
It is my great honor to be selected as the representative graduate student of the Data Science Program at National Taiwan University. I want to express my gratitude to my two thesis advisors, my family for their support, and my girlfriend for her companionship.
In my view, the future will be a world constructed by highly-intertwined information, and any organization or enterprise that can best utilize this information will seize the advantage in competition. Therefore, during my master's program, I have been highly involved in AI technologies, explicitly processing large social media data.
Congratulations to all the graduating students! I wish you continuous growth, prosperity, and a bright and wonderful future.
From Thesis to Impact: Collaboration with Cathay Financial Holdings
Thrilled to share that our research collaboration with Cathay Financial Holdings has been completed today. As a key member of the research team, we achieved three major breakthroughs:
- Demonstrating that models trained on network datasets outperform those trained on tabular datasets regarding prediction accuracy.
- Developing methodologies to address the challenges in predicting outcomes for minority groups due to inherent data distribution issues.
- In my master's thesis, I further enhanced existing models' performance, training time, and inference costs through the proposed subgraph sampling technique.
Carrefour Online Marketplace Next Purchase Prediction – 2nd Place
I'm proud that our team achieved second place in the team competition at the 2022 Data Station - Carrefour Track, co-hosted by the Ministry of Digital Affairs, R.O.C. and the Institute for Information Industry (III).
We developed a predictive model for Taiwan Carrefour's online marketplace to forecast customers' next purchases. Our team stood out among 100+ teams in the preliminary round and secured second place among 10 competitors in the finals. In the competition, we leveraged the GraphSAGE model to construct a user-item network and applied spatial-temporal techniques through edge classification to predict a consumer's potential product list for their next shopping session.