Achieving Extreme Throughput for DNN-Based In-Network Processing
Neural Networks are playing a key role in an increasing amount of application scenarios. However, their computational complexity and memory demands are challenging which limits their deployment in particular within high throughput domains such as communications.
During this talk, we will discuss an example from the space of Deep Network Intrusion Detection which leverages DNNs to identify malicious traffic for increased accuracy. We show, how through extreme specialization of compute architectures, we can effectively scale DNN inference to 100s of millions of requests/second. This is sufficient to classify traffic at up to 400Gbps wire speed, even for compute-intensive applications such as DNN inference.
Michaela Blott is a Distinguished Engineer at Xilinx Research in Dublin, Ireland, where she heads a team of international scientists driving exciting research to define new application domains for Xilinx devices, such as machine learning, in both embedded and hyperscale deployments. She brings over 25 years of leading-edge computer architecture and advanced FPGA and board design, in research institutions (ETH Zurich and Bell Labs) and development organizations. Michaela Blott is heavily involved with the international research community serving as technical program chair, organizer and technical program committee member for numerous conferences (FPL, ISFPGA, DATE, etc.). Over the years, she received several awards, most recently the Women in Tech Award 2019.