Скачать Heuristics: Intelligent Search Strategies for Computer Problem Solving: Judea Pearl бесплатно
26 июня 2009 | Автор: Admin | Рубрика: Компьютерная литература » Програм-ние и разработка » Программирование | Комментариев: 0
Heuristics: Intelligent Search Strategies for Computer Problem Solving: Judea Pearl
Addison-Wesley Pub (Sd) | ISBN: 0201055945 | 1984-04 | djvu (ocr) | 399 pages | 3.66 Mb
The Addison-Wesley series in artificial intelligence
This book is about heuristics, popularly known as rules of thumb, educated guesses, intuitive judgments or simply common sense. In more precise terms, heuristics stand for strategies using readily accessible though loosely applicable information to control problem-solving processes in human beings and machine. This book presents an analysis of the nature and the power of typical heuristic methods, primarily those used in artificial intelligence (AI) and operations research (OR) to solve problems of search, reasoning, planning and optimization on digital machines.
The discussions in this book follow a three-phase pattern: Presentation, characterization, and evaluation. We first present a set of general-purpose problem-solving strategies guided by heuristic information (Chapters 1 and 2), then highlight the general principles and properties that characterize this set (Chapters 3 and 8) and, finally, we present mathematical analyses of the performances of these strategies in several well-structured domains (Chapters 5, 6, 7, 9, and 10). Some psychological aspects of how people discover and use heuristics are discussed briefly in Chapters 1 and 4. The original intention in writing this book was to provide a cohesive and hospitable package for the theoretical results obtained at the UCLA Cognitive Systems Laboratory in the past three years. Some of these results were scattered in various reports, proceedings, and archival journals, and others buried in notebooks awaiting a sufficient incentive for refinement and publication. The compilation of this body of research into a single volume under a consistent notation and a unified logical thread now offers readers easier access to the main results and a better opportunity to assess their range of applicability. These original works are covered in Chapters 5, 6, 7, 9 and 10 and the last sections of Chapters 3 and 8.
Under the persuasive influence of colleagues and students, I later broadened the scope of the book to include some introductory material: an overview of heuristics in typical problem-solving situations (Chapter 1), a description and taxonomy of the basic heuristic search strategies used in AI (Chapters 2 and 8), a formal exposition of their main properties (Chapter 3), and a general discussion on the nature of heuristics (Chapter 4).
As it now stands, the book can fulfill three roles - it can serve as a monograph, as a reference, or as a textbook. The researcher or practitioner familiar with the standard literature on heuristic search can skip the introductory sections and turn directly to the theoretical topics, which are relatively selfcontained. Casual readers, curious enough to gain an understanding of the type of problems and techniques emerging from the AI brewery, are advised to follow the introductory material at their own pace and select the advanced reading which best matches their tastes, backgrounds, and ultimate objectives. Special care was taken to adhere to traditional mathematical notation (avoiding programming-based jargons), thereby ensuring that readers from engineering, operations research, mathematics, or computer science will find the presentation familiar and comfortable. Finally, as a textbook, the book can be used as a reference for an introductory course in AI, and as a full text in a more advanced, graduate-level class on AI control strategies or the analysis of algorithms. At UCLA, for example, we have used Chapters 1, 2, 3 (Section 3.1), and 8 to cover roughly half of the first graduate course in artificial intelligence. The other chapters are used as a text in a graduate course on heuristic algorithms that serves the curriculum of two major fields: machine intelligence and computer science theory. These chapters should also be ideal for an operations research class that focuses on the taxonomy and analysis of combinatorial search techniques.