BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Department of Electrical and Computer Engineering (HKUECE) 電機與計算機工程系 - ECPv6.16.2//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://ece.hku.hk
X-WR-CALDESC:Events for Department of Electrical and Computer Engineering (HKUECE) 電機與計算機工程系
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:20250101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Hong_Kong:20260518T160000
DTEND;TZID=Asia/Hong_Kong:20260518T170000
DTSTAMP:20260515T065207
CREATED:20260513T040826Z
LAST-MODIFIED:20260513T062829Z
UID:115975-1779120000-1779123600@ece.hku.hk
SUMMARY:RPG Seminar – Robust Multivariate Autoregressive Model Estimation under Impulsive Noise
DESCRIPTION:Zoom Link \nhttps://hku.zoom.us/j/9995553636?omn=96185323037 \nAbstract\nThis seminar presents robust multivariate autoregressive (MVAR) model estimation under impulsive noise\, with a focus on extending bias-compensated instrumental-variable methods to contaminated multichannel observations. MVAR models are widely used in time-series prediction\, system identification\, biomedical signal analysis\, and sensor-array processing. However\, conventional estimators can suffer from systematic bias under correlated measurement noise\, and their performance can degrade severely when rare but large-amplitude impulsive samples dominate covariance and correlation statistics. \nThe talk first reviews the transition from AR to MVAR modeling and explains how measurement noise enters the estimation problem. It then introduces the extended instrumental-variable bias-compensation (EIV-BC) framework for correlated noise and discusses why impulsive contamination presents an additional challenge. A robust batch EIV-BC strategy is presented\, using bidirectional estimation and residual-based outlier exclusion to identify unreliable samples before recomputing model statistics. \nExperiments are conducted on both simulated MVAR signals and real-world UCI air-quality sensor data. The results show that impulsive noise significantly degrades standard EIV-BC prediction\, while the robust extension provides more stable estimates and lower prediction errors on non-impulsive test samples. The seminar demonstrates how robust sample selection can improve MVAR estimation reliability in practical noisy sensing environments. \nSpeaker\nMr Mingxi LYU\nDepartment of Electrical and Computer Engineering\nThe University of Hong Kong \nBiography of the Speaker\nMingxi Lyu received the B.S. degree in Mechanical Engineering from Xi’an Jiao tong University in 2019\, and the M.S. degree in Mechanical Engineering from Xi’an Jiao tong University in 2022. He is currently pursuing the Ph.D. degree in electrical and electronic engineering at the Department of Electrical and Electronic Engineering\, The University of Hong Kong. His current research interests include multivariate autoregressive regression and statistical robust estimation of chirp signal with their applications. \nOrganiser\nProf. Shing Chow CHAN \nDepartment of Electrical and Computer Engineering\, The University of Hong Kong \nAll are welcome.
URL:https://ece.hku.hk/events/20260518/
CATEGORIES:Seminar
ATTACH;FMTTYPE=image/jpeg:https://ece.hku.hk/wp-content/uploads/2024/11/rpg-seminar.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Hong_Kong:20260519T160000
DTEND;TZID=Asia/Hong_Kong:20260519T170000
DTSTAMP:20260515T065207
CREATED:20260420T064633Z
LAST-MODIFIED:20260420T064633Z
UID:115720-1779206400-1779210000@ece.hku.hk
SUMMARY:RPG Seminar – From Understanding to Intervention: Interpretability-Guided Methods for Improving Large Language Models
DESCRIPTION:  \nAbstract\nLarge language models have achieved impressive performance\, but improving them efficiently and reliably requires more than scaling alone. In this talk\, I present a series of works that explore how internal understanding of LLMs can be translated into practical interventions for better capability\, efficiency\, and controllability. I begin with actionable mechanistic interpretability\, introducing a unified “Locate\, Steer\, and Improve” perspective that turns model analysis into a framework for intervention. I then show how this perspective supports several concrete advances: data-free mixed-precision quantization guided by numerical and structural sensitivity\, multilingual capability enhancement through representation shifting and contrastive alignment\, personalized multi-teacher distillation that routes each prompt to its most suitable teacher\, and coarse-to-fine selective fine-tuning for mitigating catastrophic forgetting while preserving general versatility. Together\, these works reflect a common theme: interpretability is not only a tool for explaining LLMs\, but also a principled basis for designing more efficient training\, compression\, and adaptation methods. \nSpeaker\nMr Hengyuan ZHANG\nDepartment of Electrical and Computer Engineering\nThe University of Hong Kong \nBiography of the Speaker\nHengyuan Zhang is a Ph.D. candidate at the University of Hong Kong\, supervised by Prof. Ngai Wong and Prof. Hayden Kwok-Hay So. His research focuses on the improvement of efficiency and interpretability within large language models. He aims to uncover and characterize the internal processes that govern model behavior\, with the goal of improving model speciality\, interpretability\, and reliability in real-world deployments. He has published multiple papers in leading venues such as ACL\, EMNLP\, TKDD\, and NeurIPS. \nOrganiser\nProf. Ngai WONG \nDepartment of Electrical and Computer Engineering\, The University of Hong Kong \nAll are welcome.
URL:https://ece.hku.hk/events/20260519/
LOCATION:Room CB-603\, 6/F\, Chow Yei Ching Building\, The University of Hong Kong
CATEGORIES:Seminar
ATTACH;FMTTYPE=image/jpeg:https://ece.hku.hk/wp-content/uploads/2024/11/rpg-seminar.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Hong_Kong:20260520T110000
DTEND;TZID=Asia/Hong_Kong:20260520T120000
DTSTAMP:20260515T065207
CREATED:20260514T081134Z
LAST-MODIFIED:20260514T081134Z
UID:116062-1779274800-1779278400@ece.hku.hk
SUMMARY:RPG Seminar – Foundation-style Methods for Real-Time Statistical Dependency Measurement and Its Applications
DESCRIPTION:Zoom Link \nhttps://hku.zoom.us/j/9021481973?omn=94827890905 \nAbstract\nMutual information has long served as a principled measure of statistical dependence\, but computing it from empirical samples is notoriously difficult: neural estimators rely on costly gradient-based optimization for every new dataset\, which limits their use inside real-time and large-scale pipelines. This talk presents a framework that pretrains a neural estimator on a synthetic meta-distribution and then evaluates any new distribution in a single forward pass — turning a per-dataset optimization problem into an inference problem. Cheap\, online dependency readings further act as many applications\, including live training diagnostics\, and as a regularization signal for detecting key frames. \nSpeaker\nMr. Zhengyang HU\nDepartment of Electrical and Computer Engineering\nThe University of Hong Kong \nBiography of the Speaker\nZhengyang Hu\, a three-year PhD in ECE\, mainly focuses on statistical dependency measurement and foundational-style data science models. \nOrganiser\nProf Yanchao YANG \nDepartment of Electrical and Computer Engineering\, The University of Hong Kong \nAll are welcome.
URL:https://ece.hku.hk/events/20260520/
CATEGORIES:Seminar
ATTACH;FMTTYPE=image/jpeg:https://ece.hku.hk/wp-content/uploads/2024/11/rpg-seminar.jpg
END:VEVENT
END:VCALENDAR