威廉希尔足球官网-学术报告

报告题目：Adversarial Attacks in Deep-Learning-Based Image Forensics

报告人： 王真 教授

报告时间：2021年12月9日（周四）上午10:00

腾讯会议：455 562 114

主办单位：地下空间智能控制教育部工程研究中心

williamhill威廉希尔官网

报告简介

In the new AI era (or equivalently in the “Fake News” era), seeing will no longer be believing and even truth will not be believed. People can easily manipulate digital images or generate highly visually convincing fake images and videos, and therefore pose critical challenges in digital image forensics (DIF) analysis. Deep learning has been widely employed for DIF tasks. Unfortunately, both digital images and deep learning models are vulnerable to manipulations and attacks, intentionally or unintentionally. This talk gives a brief review of my group’s recent research efforts in the intersection of deep learning and DIF -- adversarial deep learning, with focus on adversarial attacks. We consider potential adversarial attacks and explore novel approaches to study vulnerabilities of deep learning models, by investigating three fundamental learning tasks: matching, classification and regression. We present novel attacks (both in the digital domain and physical domain) for several essential models (e.g., fake face imagery forensics; camera-LIDAR 3d object detection; and single object tracking in videos). Potential future directions in achieving more trusting deep learning solutions for trusting digital images will also be discussed.

个人简历

王真，加拿大英属哥伦比亚大学（UBC）电子与计算机工程系终身教授、IEEE Fellow。她于清华大学电机系获学士学位，美国康涅狄格州立大学（UConn）获硕士和博士学位，美国马里兰大学博士后。她从2004年起任教UBC，领导数字信号图像处理、机器学习和生物工程实验室。在权威杂志上发表160余篇期刊论文及100余篇国际会议论文。曾担任IEEE多个期刊的编辑，现任 IEEE Signal Processing Letters 的主编，多次参与组织领域内的顶级IEEE会议，担任大会主席、学术委员会主席等。