Networks are becoming an essential part of the infrastructure. Over the last decades, effective intrusion detection techniques have been developed to detect and prevent cyber-attacks from networks. Future networks will evolve to have higher capacity, lower latency, and vast number of connected devices, facilitating increased and more sophisticated intrusion attacks. Machine learning based approaches have shown high potential to address these challenges, however, many of them would suffer from data scarcity and imbalance problems if they are applied in intrusion attacks.
This talk will introduce the general background of machine learning approaches for network intrusion detection, and discuss the challenges of data scarcity and imbalance and their impacts to detection performance. The talk will show how novel deep adversarial learning and statistical learning methods could augment attack samples and deep learning models could be trained to address emerging intrusion attacks. Some preliminary experiment results will be presented to demonstrate the advantages and limitations of the proposed methods and facilitate further discussion.


To attend via Webex, please register at: https://tinyurl.com/ChunboLuoWebex

Any questions, please contact Sarah Mackenzie or Selina Wang