The Applications of Deep Learning on Traffic Identification

Explore the cutting-edge applications of deep learning in network traffic identification through this 26-minute Black Hat conference talk. Delve into the limitations of traditional feature-based identification methods and discover how neural networks and deep learning can revolutionize protocol detection. Learn about automatic feature learning, protocol identification, and anomalous protocol detection techniques. Gain insights into the implementation details, including input layers, regression, and evaluation processes. Understand the power of parallel computing in this context and see real-world examples of feature distribution and application. Conclude with a discussion on the effectiveness and potential future developments of this innovative approach to traffic identification.