Introduction to artificial intelligence. Translated from the German by Nathanael Black. (Q625106)

This accessible and concise introduction to the field of artificial intelligence (AI) is intended primarily for self-study or as a foundation of a short course on the subject. The book consists of ten topic chapters, each of which offers an extended list of exercises. Chapter 11 contains solutions to all exercises. Additional teaching resources, including lecture slides, are available on the book's website. Chapter 1 covers multiple introductory topics, such as definitions of the field, history of AI, different types of AI agents, and knowledge-based systems. Chapter 2 is an introduction to propositional logic. It discusses syntax, semantics, and proof theory, and briefly addresses complexity issues. Special attention is given to the resolution method. Chapter 3 deals with first-order logic describing its syntax, semantics, and proof theory. It further discusses the resolution method and outlines several resolution strategies. Mathematical examples are used to demonstrate automated theorem proving. Limitations of mathematical logic are discussed in Chapter 4, where a brief introduction to non-classical logics (fuzzy logic and probabilistic logic) is provided. Chapter 5 introduces logic programming and Prolog by means of simple examples. Self-modifying programs and constraint logic programming is also discussed. AI search is covered in Chapter 6. It encompasses uninformed search, heuristic search, informed search, and different searches in games with opponents. Reasoning with uncertainty is covered in a pretty detailed way in Chapter 7. After introducing the basics of computing with probabilities, conditional probability, and the Bayes theorem, the author discusses the principle of maximum entropy and an inference rule for reasoning with probabilistic knowledge. Special attention is paid to Bayesian networks and MaxEnt-Systems. Lexmed, an expert system for diagnosing appendicitis, is used as an example to illustrate the MaxEnt method and to show how it can be used in knowledge discovery. Chapter 8 discusses machine learning and data mining. Several supervised learning methods, including decision tree learning, learning of Bayesian networks, case-based learning, and the nearest neighbor method are thoroughly covered. Chapter 9 is devoted to neural networks. It outlines the historical progression of the field by defining a model of the neuron and its interconnectivity, the Hopfield model, and the back-propagation algorithm. It briefly discusses NETtalk, a neural network capable of learning to speak, as well as support vector machines. Chapter 10 is devoted to reinforcement learning and dynamic programming.