研讨会信息
🎤 Speaker: Prof. Yatao BIAN
Assistant Professor (tenure-track),
Department of Computer Science, National University of Singapore (NUS)
(ybian@nus.edu.sg)
📰 Title: From Pretraining to LLMs: Tackling Label Scarcity in Molecular Graphs
⏰ Time: 1:30-2:20 PM, BJ Time
📆 Date: 31 Oct., 2025 (Fri.)
📍 Venue: Hall C, GZ Campus
Online Zoom link:
https://hkust-gz-edu-cn.zoom.us/j/96983408548?pwd=moZgiftuonqonX6aTbOjTPDbkKRCWe.1
Zoom ID: 969 8340 8548
Passcode: 823262
研讨会概要
Abstract
The rapid progress of machine learning is reshaping molecular science, a core pillar of the natural sciences. Yet, practical molecular discovery is often bottlenecked by label scarcity. This talk presents methods for molecular graph learning that remain robust in low-data regimes. I will begin by introducing the development of molecular graph foundational models designed to mitigate the label scarcity problem, including a pretrain-finetune paradigm and a powerful graph Transformer architecture that significantly enhance model performance in low-data regimes. Building on this foundation, I will demonstrate how to extend this paradigm into the era of large language models (LLMs) by aligning text and structure, integrating the power of LLMs with the intricate structure of molecules.
分享者简介
Prof. Yatao BIAN
Assistant Professor (tenure-track),
Department of Computer Science,
National University of Singapore (NUS)
Yatao Bian is a tenure-track assistant professor of computer science with the National University of Singapore. He received the PhD degree from the Institute for Machine Learning, ETH Zurich. From 2015 to 2020, he was an associate fellow with the Max Planck ETH Center for Learning Systems. He is dedicated to expanding the frontiers of AI capabilities in scientific intelligence, driving transformative technological advancements to address the most critical challenges in scientific research. His research interests include reasoning-empowered foundation models, LLM/agent for science, graph machine learning and energy based learning. He has authored or coauthored several papers on machine learning top conferences/journals such as NeurIPS, ICML, ICLR and T-PAMI. He is the area chair of NeurIPS and ICLR. He was a reviewer/PC for conferences, such as ICML, NeurIPS, ICLR, CVPR, AAAI, STOC and journals like JMLR, T-PAMI and Nature Machine Intelligence.
扫描加关注
获取更多AI学域消息
SCAN & FOLLOW US!