Secured Machine Learning Engineering


Over the past decades, deep learning (DL) systems achieved tremendous success and gained great popularity in various cutting-edge applications, e.g., robotics, image processing, speech processing, and medical diagnostics. A deep neural network (DNN), as a type of deep learning systems, is the key driving force behind its recent success. However, the security and quality assurance techniques for DL are still as the early stage, and a plethora of studies have shown that the state-of-the-art DL systems suffer from various vulnerabilities which can lead to severe consequences when applied to real-world applications. We believe that obtaining robust deep learning system of the next generation does not solely reply on foundational research of deep learning, but also demands huge engineering effort as well, which we define as “Secured Machine (Deep) Learning Engineering”. We have conducted consecutive work along the general purpose security and quality assurance of DL engineering, we will highlight several of our very recent exciting achievements in this direction and their potential immediate potential impact on large scaled and industry level applications.