IDSSP LAB

09/07/2026

Class of 2026 🎓

Congratulations to our graduates! May your future be bright, and may all your dreams come true. We are proud of you!

07/07/2026

Welcome our new iDSSPals ✨

We're excited to have you join our lab and become part of the iDSSP family! 🎉

A new journey of learning, collaboration, and discovery begins here. We can't wait to see the ideas you'll bring, the challenges you'll tackle, and the memories we'll create together.🥰

03/07/2026

Congratulations to Van Quang Nghiem, Si-Qin Lyu, and Professor Che Lin on their paper being Early Accepted to **Medical Image Computing and Computer Assisted Intervention (MICCAI) 2026**! 🧠✨

📄 Paper: “ReMiX-Seg: Latent Modality Completion Meets Expert Routing for Robust Glioma Segmentation”

This work originated from the final project of the Artificial Intelligence and Intelligent Medicine (AIIM) course and was further developed into an early-accepted paper at MICCAI 2026, one of the leading conferences in medical image computing and computer-assisted intervention. This achievement also reflects the strong research-oriented training and high-quality learning environment provided by the course.

With an overall acceptance rate of 27.5% and an early acceptance rate of only 9%, MICCAI 2026 is highly competitive, making this accomplishment especially meaningful.

Huge congratulations to the team on this outstanding achievement! 👏🚀

02/07/2026

🎓 Graduation Party for our lab members 🥳

Congratulations on your graduation! 🎉
Wishing you all the best as you begin this exciting new chapter. May your future be filled with happiness, success, and endless opportunities. ❤️✨

18/06/2026

很高興與大家分享，我們最新的摘要 "Amplifying the predictive power of bulk transcriptomics through computational deconvolution" 已被 Multi-Omics 2026 錄取，將以海報發表（Poster Presentation）的形式展出！

我們的研究聚焦於如何從大量轉錄組學（Bulk transcriptomics）數據中挖掘更深層的洞察，非常期待能在現場分享我們的成果並與學界、業界的夥伴交流。

歡迎點擊連結查看完整議程：https://www.multiomics2026.org/

Excited to share that our latest abstract, "Amplifying the predictive power of bulk transcriptomics through computational deconvolution", has been accepted for a poster presentation at Multi-Omics 2026!

Our work focuses on unlocking deeper insights from bulk data, and we cannot wait to share our findings and connect with the community.

Check out the full program here: https://www.multiomics2026.org/

REGISTER NOW Invitation to multi-omics 2026 - Brisbane, Australia Meeting start: 08:00 Sunday 6 DecemberMeeting end: 20:00 Tuesday 8 December 2026On behalf of the organising committee, we are delighted to invite you to join us at Multi-Omics 2026, held at the Brisbane Convention & Exhibition Centre,...

11/06/2026

🎉 我們的 FinTech 課程圓滿落幕啦！ 🚀

真的非常為所有同學們感到驕傲！大家在期末專題中展現了超乎想像的努力與創意，今天的成果發表會精彩萬分、亮點滿滿，看著台上的你們閃閃發光，真的非常感動！💡✨

恭喜所有獲獎的團隊 🏆 每一份榮譽都是你們實至名歸的肯定！

同時，也要由衷感謝百忙之中蒞臨的超強評審陣容。謝謝您們用業界與學術界的精準眼光，給予同學們最專業的指導與精彩點評：
✨ Spencer Huang (禾多移動 董事長)
✨ Alex Chia (群益期貨 董事長)
✨ Edward Mao (群益期貨 總經理)
✨ 黃維本 (群益期貨 高級執行副總經理)
✨ 王釧茹 教授 & 王志宇 教授 (中央研究院)

特別感謝 @群益期貨 與 .ai 禾多移動 慷裁贊助本次的專題獎金，有您們的大力支持，讓學生的創新想法得以走得更遠 🙏✨

非常享受這學期與大家一起探索金融科技的旅程，希望每位同學都收穫了滿滿的知識與回憶！未來讓我們繼續一起成長、創新、大步向前！💙

🎉 What an amazing conclusion to this year’s FinTech course! 🚀

We are incredibly proud of all our students! Your hard work, dedication, and late nights truly shined through in your brilliant final project presentations today. The level of innovation and ex*****on was absolutely outstanding! 💡✨

A huge congratulations to the winning teams who received project awards—you absolutely earned it! 🏆

Endless gratitude to our phenomenal panel of judges for sharing their invaluable insights and expertise with our students:
✨ Spencer Huang (Chairman, Avivid)
✨ Alex Chia (Chairman, Capital Future)
✨ Edward Mao (President, Capital Future)
✨ Wei-Pen Huang (Senior Executive Vice President, Capital Future)
✨ Dr. Chuan-Ju Wang & Chih-Yu Wang (Academia Sinica)

A special shout-out and heartfelt thanks to Future and .ai for generously sponsoring our prize awards! Your support makes a world of difference to these young innovators 🙏💙

It has been an absolute privilege teaching this course, and we hope it was a deeply rewarding journey for every single one of you. Here’s to dreaming big, building smart, and continuing to innovate together in the world of FinTech! 🌟

#金融科技 #成果發表 #期末專題 #國立臺灣大學

03/06/2026

IDSSP 林澤教授受邀至建中，分享「虛擬病人平台建立與醫學教育」的發展經驗，與同學們一起探討 AI 時代下醫學、資訊與資料科學跨領域合作的未來可能。

這次也特別帶著資料科學博士生，以及臺大醫學系與臺大資工系的年輕同學們一同回到校園，希望透過不同背景學生的交流，鼓勵大家勇敢跨出舒適圈，在未來面對 AI 浪潮時，不只是被取代，而是學會如何與 AI 協作、整合人文與科技，找到屬於自己的價值與定位。

Professor Lin Che from IDSSP is invited to Taipei Municipal Chien Kuo High School to share insights on the development of virtual patient platforms and the future of medical education in the AI era.

Together with PhD students in Data Science, as well as young students from National Taiwan University College of Medicine and the Department of Computer Science, we returned to campus to encourage interdisciplinary collaboration among future generations.

Through conversations across medicine, computer science, and AI, we hoped to inspire students not only to face the challenges brought by artificial intelligence, but also to learn how to collaborate with AI, integrate humanity with technology, and discover their unique value in a rapidly changing world.

02/06/2026

✨ InnoVEX 2026 現場：以 AI 推進電子商務推薦 🚀

我們很高興能於 InnoVEX 2026 展出我們的技術：
大語言模型強化之商品標註與多模態推薦整合系統

我們的系統由兩個核心模組組成：

1️⃣ BETag：利用使用者歷史紀錄微調大型語言模型，生成更能反映使用者行為的商品標籤，支援可擴展且低延遲的離線部署。

2️⃣ MTSTRec：透過 Time-aligned Shared Token（TST）與 Transformer 架構，整合 ID、文字、價格與圖片風格資訊，以更精準地建模使用者偏好。

🔬 透過結合 BETag 的行為增強標籤生成與 MTSTRec 的時間對齊多模態融合，我們的系統能提升檢索與推薦效能。這兩個模組也都設計為可輕鬆整合至現有的電子商務平台中。

歡迎大家在 InnoVEX 2026 期間來我們的攤位參觀，進一步了解我們在標籤生成與推薦系統上的技術與應用！

📍 地點：南港展覽館2號館4樓 S1204 攤位
📅 日期：2026 年 6 月 2–5 日

✨ Live at InnoVEX 2026: Advancing E-Commerce Recommendation with AI 🚀

We are excited to be exhibiting at InnoVEX 2026 and showcasing our technology:
LLM-enhanced Unified Tagging and Multimodal Recommendation System

Our system is built around two core modules:

1️⃣ BETag: Generates behavior-enhanced item tags using an LLM fine-tuned on user history, enabling scalable and low-latency offline deployment.

2️⃣ MTSTRec: Integrates ID, text, price, and image style information through a Time-aligned Shared Token (TST) and Transformer architecture to better model user preferences.

🔬 By combining BETag’s behavior-enhanced tagging with MTSTRec’s time-aligned multimodal fusion, our system delivers strong retrieval and recommendation performance. Both modules are designed to be easily integrated into existing e-commerce platforms.

If you’re around, stop by our booth and chat with us about tagging and recommendation systems!

📍 Location: Booth S1204, 4F, TaiNEX 2
📅 Date: June 2–5, 2026

29/05/2026

🎉 Congratulations to our lab member, **Van Quang Nghiem (Andrew)**, a first-year Master’s student in GICE, for being selected to give an oral presentation at the AI EECS Consortium Technology Showcase held at Barry Lam Hall.

The event brought together **6 Elite Scholarship Master’s students** and **14 Elite PhD students** from EECS for oral presentations, poster sessions, and exchanges with partner companies. Andrew shared his current research in medical imaging during the oral session.

28/05/2026

【你的平台推薦內容，有打中使用者需求嗎❓】
​
由 國立臺灣大學 National Taiwan University #林澤 教授團隊開發的「 #大語言模型強化之商品標註與多模態推薦整合系統」，結合大型語言模型與多模態 AI，讓推薦系統不只看購買紀錄，更能理解使用者真正的興趣與偏好
​
此技術可應用於電商、影音與數位平台，提升推薦精準度、轉換率與用戶黏著度，並榮獲未來科技獎及產業合作的商機
​
🔎想知道 AI 如何讓推薦更懂人？點擊連結看更多
▍查看技術清單：https://pse.is/958mfl
▍預約媒合交流：https://pse.is/958mk2