David Zane was born and raised in Honolulu, Hawaii. He is currently an undergraduate student at Northwestern University in Evanston, IL pursuing a BS in Computer Science. He enjoys programming algorithms and new interfaces. Outside of class David enjoys traveling, sports, spending time with family & friends, and eating lots of fresh broccoli.
Home Island: Oahu
Institution when accepted: Northwestern University
Akamai Project: Deep Learning for Network Anomaly Detection: Building BiGAN
Project Site: Maui High Performance Computing Center – Kihei, Maui
Mentors: Dr. Wesley Emeneker, Dr. Robert Trevino
Collaborator: Sherie Yip
The United States Department of Defense (DoD) operates networks around the world which are constantly under attack. These attacks must be identified and defended against to ensure the safety of the United States and its people. Current cybersecurity software, specifically network anomaly detection, is able to detect known anomaly signatures but cannot learn to identify new attacks without human assistance. My project aims to develop an improved method of identifying potential attacks by using a deep learning Bidirectional Generative Adversarial Network (BiGAN). A BiGAN model consists of two competing neural networks called the generator and discriminator which attempt to outsmart each other. The generator attempts to create fake data similar to actual data while the discriminator learns to discern fake from real data. After training both components concurrently, the discriminator can then be used to effectively identify abnormal data. The KDD99 dataset, a popular network traffic dataset used for machine learning cybersecurity, was used to train and test the model. Results of BiGAN anomaly detection are forthcoming. This project serves as an exploration into how deep learning can be used for cybersecurity and will hopefully inform future development of the technology.