AI&DSA研讨会 | Explainable Time series Learning: Method and Evaluation

我爱免费

作者：微信文章






研讨会信息



🎤 Speaker: Dr. Dongsheng Luo,

Assistant Professor, Knight Foundation School of Computing and Information Sciences, Florida International University;

Ph.D. in Information Science and Technology from Pennsylvania State University

📰 Title: Explainable Time series Learning: Method and Evaluation

⏰ Time: 9:30–10:30 AM,

📆 Date: Mar. 5, 2026 (Thursday)
📍 Venue: E1-202, GZ CampusOnline Zoom Meeting link:
https://hkust-gz-edu-cn.zoom.us/j/92708037211?pwd=RdIFhhB6R544qRCIfefupQm7YQlbjy.1

Meeting ID: 927 0803 7211

Passcode: ait

研讨会概要



Abstract

While time series data drives critical decision-making in domains from healthcare to finance, modern predictive models often function as opaque black boxes. This limits our ability to understand underlying temporal mechanisms and prevents deployment in high-stakes environments. My research addresses this critical gap by building trustworthy AI systems with explainability at their core. In this talk, I will present a comprehensive framework for explainable time series learning, focusing on both novel methodology and rigorous evaluation. I will introduce TimeX++, a method that captures temporal dynamics by learning explanations through information bottleneck principles, and F-Fidelity, a robust framework designed to rigorously evaluate the faithfulness of these explanations. Finally, I will outline my future research vision for advancing Trustworthy AI, exploring how explanation-assisted learning and human-AI interaction can bridge the gap between complex temporal models—including emerging foundational architectures—and human decision-makers.

分享者简介





Dr. Dongsheng Luo,

Assistant Professor,

Knight Foundation School of Computing and Information Sciences,

Florida International University

Dongsheng Luo received his Ph.D. in Information Science and Technology from Pennsylvania State University in 2022 and his B.Eng. in Computer Science from Beihang University in 2017. His research focuses on trustworthy AI for scientific discovery, with emphasis on explainable AI methods for scientific data including graphs and time series. He has published in top venues including NeurIPS, ICML, ICLR, KDD, and TPAMI. He has served as Area Chair for ICLR, NeurIPS, and other major conferences.



扫描加关注

获取更多AI学域消息

SCAN & FOLLOW US!