Safe And Robust Machine Learning

Deploying machine learning in the real world is a lot different than developing and testing it in a lab. Quenton Hall, AI systems architect at Xilinx, examines security implications on both the inferencing and training side, the potential for disruptions to accuracy, and how accessible these models and algorithms will be when they are used at the edge and in the cloud. This involves everything from speed of detection and remediation to the tradeoffs involving safety, how to verify models are accurate and secure, and where those attacks can occur in the training-to-inferencing flow.