Due to the common practice of using pre-trained weights in deep learning, backdoor attacks have become a security threat to AI systems. A backdoor model is a network that can work genuinely on clean data but misbehaves when a specific trigger pattern appears. In this episode of VinAI Research Seminar Series, let’s discuss the topic of “Backdoor Attacks and Defenses” with our research scientist – Dr Tuan Anh Tran. In this tutorial, he will first review the attack methods from simple to complex. Then, he will discuss the proposed techniques for both model defense and run-time defense.
This is an online seminar and will be streaming on our facebook fanpage VinAI Research. Let’s save the date and join us: 10.00 am – 11.30 am, Tuesday, 21/07/2020
Prior to VinAI, Dr Tuan Anh Tran was an Applied Scientist at Amazon Rekognition, working on facial image processing APIs. He received Bachelor of Engineering from the Hanoi University of Science and Technology in 2010 and a Ph.D. in Computer Science from University of Southern California in 2017, working with Professor Gerard Medioni. His research interests are in computer vision, particularly in facial image analysis. He has received several honors including Vietnam Talents 2010, Imagine Cup Vietnam 2009, and Vietnam Education Foundation fellowship 2012.