ABSTRACT
This chapter reviews a main area of sub-symbolic artificial intelligence (AI) that underlies most of the recent advancements which have brought AI to public attention. It describes a workhorse for machine learning that has contributed strongly to a first major wave of machine learning applications in industrial settings that of support vector machines. The chapter discusses the probabilistic setting of machine learning and its relation to Bayesian methods. Machine learning is about describing data to build models that can be used to predict previously unseen data. A classic example is recognizing handwritten characters or even just digits. Image recognition is the type of application to which deep learning has made considerable contributions and revolutionized computer vision. Convolution can be applied to a neural network by replacing the previous matrix operations with the convolution operation. The chapter then outlines the reason for new exciting advancements in deep learning's applications, such as image processing and natural language processing.
