Deep Neural Networks have become the most powerful software development technique in recent years, leapfrogging other more established, but increasingly obsolete, artificial intelligence techniques. They are responsible for most of the recent wave of successful AI and Machine Learning applications for image and speech recognition, natural language, big data analytics and even deep fake videos. At the same time, over-anthropomorphized explanations invoke human notions of “learning” or “neurons” to try to explain the technology and lead to unfounded fears of synthetic intelligences running amok on our streets, in our homes and on our battlefields. Just as systems engineers need a sufficient understanding of electrical engineering, mechanical engineering and software engineering, they must come to understand artificial intelligence as a new engineering discipline. While many courses are available for AI specialists and programmers, this tutorial is designed for systems engineers and requires no programming background or specialized mathematical knowledge.
Part I of the tutorial provides an overview of the field of Artificial Intelligence. Part II focuses on deep neural networks, starting from first principles and showing how they work—taking all the mystery out of important concepts like multi-layered neural networks, forward and back propagation, hyperparameter tuning and training data. Part III covers applications like convolutional neural networks for image recognition, recurrent neural networks for machine translation, word embeddings for natural language processing, reinforcement learning for physical systems control, and will provide an introduction to Explainable AI. Part IV will focus on the relationship between artificial intelligence and systems engineering in practice.