Overview of Adversarial Machine Learning (AML) with Applications to Network Anomaly Detection (NAD)

3 Mar
2021

George Kesidis and David J. Miller (Pennsylvania State University, USA)

External Presentation (External Speaker)

In this talk, we will give an overview of “adversarial” attacks on deep learning, generally known as adversarial machine learning (AML). In particular, we will describe cutting-edge defenses against backdoor/trojan and test-time evasion attacks. The seminar concludes with a discussion network-based anomaly detection (NAD) of zero-day attacks based on packet-flows. This research was funded in part by NSF and AFOSR grants and gifts from Cisco.

About George Kesidis and David J. Miller

George Kesidis and David J. Miller are full professors of EECS at Pennsylvania State University with decades of research and teaching experience in machine learning, communications and networking, security, cloud computing, and performance evaluation. Recently, their research has been funded by grants from NSF, DARPA, ONR, AFOSR and AFRL, and by several gifts from Cisco and Amazon. They are also co-founders of a small start-up working in the areas of security and machine learning. Most of our recent papers are available in technical report form at arxiv.org. Additional publication and funding details are given here: http://www.cse.psu.edu/~gik2.

This event will be conducted in English

  • Location: https://zoom.us/j/94080671779?pwd=QVBrdnJCbS8ydWM5eXlxdW9leHNidz09
  • Organization: NETCOM Research Group (Telematics Engineering Department, UC3M); IMDEA Networks Institute
  • Time: 16:30