ABSTRACT
Artificial intelligence (AI) is a field where one size definitely does not fit all, and an understanding of the different types of machine learning (ML) available to us is key to building successful solutions for autonomous networks. It reviews the relevant AI algorithms from traditional ones, such as regressions and clustering, to modern ML techniques, such as natural language processing, deep learning (DL), convolutional networks, and reinforcement learning. The chapter also describes successful applications of AI in areas of speech recognition, language translation, video processing, and facial and object recognition. It provides historical perspective on the evolution of machine intelligence and the innovative researchers involved, from early "perceptrons" through expert systems, optimization, and forecasting, down to modern DL. General optimization uses a technique called "gradient descent" (GD), which looks at the error in the current result versus the error in previous results and pushes the function "downhill".
