IDSSP LAB

🏫✨ 歡迎建國中學學生們蒞臨 iDSSP Lab 參訪！

今天很開心迎接來自建國中學的同學們到 iDSSP Lab 進行參訪交流 🤝
透過本次活動，希望同學們能更認識人工智慧、資料科學與智慧醫療等研究方向，也能對未來的學習與研究探索產生更多想像與啟發 💡🔬

謝謝同學們的熱情參與，期待大家都能帶著滿滿收穫回家，也歡迎未來有機會再次交流！🌟

#建國中學 #實驗室參訪

我們的團隊參與於倫敦舉辦的 ISBI（IEEE International Symposium on Biomedical Imaging）🌍，並於會中進行口頭報告 🎤

報告主題為「Learning from Limited Multi-Phase CT: Dual-Branch Prototype-Guided Framework for Early Recurrence Prediction in HCC」，探討在有限多期CT資料下，利用原型引導學習進行肝癌早期復發預測。

此次會議也與來自各地的研究者交流，收穫豐富 ✨

Our team participated in ISBI (IEEE International Symposium on Biomedical Imaging) held in London 🌍 and delivered an oral presentation 🎤

The presented work, titled “Learning from Limited Multi-Phase CT: Dual-Branch Prototype-Guided Framework for Early Recurrence Prediction in HCC,” focuses on leveraging prototype-guided learning for early recurrence prediction using limited multi-phase CT data.

It was a great opportunity to engage with researchers worldwide and exchange ideas ✨

🎉 Congratulations to our lab on this great achievement! 🎉

We are delighted to share that our paper, “NASH: Numerically Aware Scoring Heuristic for Robust Semantic Similarity,” has been accepted as a Findings Paper at ACL 2026!

This year, ACL received 12,148 submissions, with only 19% accepted as Main Papers and 18% as Findings Papers. We are truly proud to be among the accepted works in such a highly competitive venue.

👏 A big thank you to all the contributors for their hard work and dedication. We look forward to achieving even more exciting milestones in the future!

🎉 iDSSP Lab Year-End Party at 果然匯 🍽️

We’re proud of all the wonderful achievements we’ve accomplished together this year—thank you for everyone’s hard work and dedication! 💪✨
Let’s keep moving forward and make the coming year even brighter. 🌟

Wishing you all a joyful New Year filled with happiness, success, and good health! 🧧🥂

I am thrilled to announce that our paper, "Beyond Curated Knowledge: Structural Protein Embeddings Enhance GNN-Based Personalized Cancer Prognosis," has been accepted by the IEEE Journal of Biomedical and Health Informatics (J-BHI)! 📄✨

While most AI models predict cancer risk by looking at how much a gene is active, we took it a step further: we taught the AI to understand what the protein actually looks like.

By using Protein Language Models (think ChatGPT, but for amino acids), we captured the structural shape of cancer-driving proteins. We then combined this with patient data using Graph Neural Networks to predict 5-year survival outcomes more accurately. This approach helps identify high-risk patients who might need closer monitoring.

A massive thank you and congratulations to the team for their hard work: 👏 Sofia Ormazabal Arriagada 👏 Tsung-Wei Lin 👏 Marta Misztal 👏 Prof. Che Lin

🎉 恭喜 iDSSP Lab 論文獲 ICLR 2026 接受！ 🎉

本研究論文
📄 「Atomic HINs: Entity-Attribute Duality for Heterogeneous Graph Modeling」
已獲
🏆 International Conference on Learning Representations（ICLR 2026） 接受。

本研究提出一種全新的 schema-aware 異質圖建模方法，透過 entity–attribute duality 與 schema-refinement framework，實現對異質資訊網路更系統性且有效的建模方式 🚀

本研究為與中央研究院資訊科技創新研究中心 王志宇（Chih-Yu Wang）教授 之合作成果。誠摯感謝合作夥伴在研究過程中的寶貴討論與支持。

期待在 ICLR 2026 與大家交流！📚✨

🎉 Big congratulations to iDSSP Lab on an ICLR 2026 paper acceptance! 🎉

Our paper
📄 “Atomic HINs: Entity-Attribute Duality for Heterogeneous Graph Modeling”
has been accepted by
🏆 International Conference on Learning Representations (ICLR 2026).

This work introduces a new schema-aware heterogeneous graph modeling paradigm, enabled by entity–attribute duality and a schema-refinement framework, allowing more principled and effective modeling of heterogeneous information networks 🚀

This research was conducted in collaboration with Prof. Chih-Yu Wang at the Research Center for IT Innovation, Academia Sinica. We sincerely thank our collaborators for their insightful discussions and support throughout this work.

Looking forward to sharing and discussing this work at ICLR 2026! 📚✨

🎉 狂賀 iDSSP Lab 論文獲 ISBI 2026 接受！ 🎉

我們的研究論文
📄 「Learning from Limited Multi-Phase CT: Dual-Branch Prototype-Guided Framework for Early Recurrence Prediction in HCC」
已正式被國際頂級醫學影像會議
🏆 IEEE International Symposium on Biomedical Imaging (ISBI 2026) 接受！

本研究為 衛福部（MOHW）合作研究計畫的重要成果之一，結合臨床與人工智慧團隊，提出一套 Dual-Branch Prototype-Guided Framework，在多相位 CT 資料受限的情境下，仍能有效預測 肝細胞癌（HCC）術後早期復發（Early Recurrence），為精準醫療與多院模型泛化帶來突破性進展 🚀

論文將於 ISBI 2026 現場發表，並收錄於 IEEE Xplore 國際學術資料庫 📚

誠摯感謝所有參與本 MOHW 計畫的成員與臺大醫院（NTUH）臨床合作團隊的支持與合作。

🎉 Big congratulations to iDSSP Lab on an ISBI 2026 paper acceptance! 🎉

Our paper
📄 “Learning from Limited Multi-Phase CT: Dual-Branch Prototype-Guided Framework for Early Recurrence Prediction in HCC”
has been accepted by
🏆 IEEE International Symposium on Biomedical Imaging (ISBI 2026).

This work is a major research outcome of our MOHW (Ministry of Health and Welfare) collaborative project, integrating clinical and AI expertise to develop a Dual-Branch Prototype-Guided Framework that enables robust early recurrence prediction for hepatocellular carcinoma (HCC) even with limited multi-phase CT data 🚀

The paper will be presented on site at ISBI 2026 and published in IEEE Xplore 📚

We sincerely thank all members involved in this MOHW project and our clinical collaborators at National Taiwan University Hospital (NTUH) for their support and collaboration.

#衛福部計畫

Congratulations to our lab member, Van Quang Nghiem, for receiving the EECS Global Elite Cultivation Fellowship Scholarship and representing international students at the NTU AI–EECS Consortium 2025 Annual Appreciation Gathering.✨✨✨

恭喜林澤老師獲得電資學院114年度學術貢獻獎✨✨✨
感謝電資學院與院長的肯定👍
也感謝團隊的努力與支持，學術路上持續前行🏃

【Expert Talk】

Speaker: Professor D**g Yuan (The University of Sydney)
Topic: Design and optimise computing systems for AI: from data collection to training and inference
Date: 2025 December 12th (Fri.) 14:00 - 15:00
Venue: Barry Lam Hall R201
Language: English

🎉恭賀iDSSP團隊榮獲國科會產學成果-簡報組 特優獎！
感謝實驗室團隊的努力與合作，也謝謝評審的肯定，未來我們會繼續努力！💪✨
🏆 Congratulations to our team for receiving the Outstanding Award in the Industry-Academia Achievement – Presentation Category from the National Science and Technology Council (NSTC)!
We sincerely thank our lab members for their hard work and teamwork, as well as the judges for their recognition. We will continue to strive for excellence! 🚀