• Contribute
• What's New
• Submit News

	8.5
	Author: Stuart Russell and Peter Norvig Excerpt: Artificial Intelligence: A Modern Approach presents a unified, coherent picture of the AI field based on the idea of intelligent agents, and shows how to build them using AI methods. .: Buy at Amazon.com :. .: Buy at Amazon.co.uk :.

Artificial Intelligence: A Modern Approach has been the staple textbook for hundreds (660 according to the website!) of university-level Artificial Intelligence courses. Indeed, the textbook provides a very comphrensive look at many different fields of AI, with an emphasis on intelligent agents. Rather than break down each chapter, here is an overview of the book (taken from Summary of Contents):

• Artificial Intelligence
  1. Introduction
  2. Intelligent Agents
• Problem-Solving
  1. Solving Problems by Searching
  2. Informed Search Methods
  3. Game Playing
• Knowledge and reasoning
  1. Agents that Reason Logically
  2. First-Order Logic
  3. Building a Knowledge Base
  4. Inference in First-Order Logic
  5. Logical Reasoning Systems
• Acting Logically
  1. Planning
  2. Practical Planning
  3. Planning and Acting
• Uncertain Knowledge and Reasoning
  1. Uncertainty
  2. Probabilistic Reasoning Systems
  3. Making Simple Decisions
  4. Making Complex Decisions
• Learning
  1. Learning from Observations
  2. Learning in Neural and Belief Networks
  3. Reinforcement Learning
  4. Knowledge in Learning
• Communicating, perceiving, and acting
  1. Agents that Communicate
  2. Practical Natural Language Processing
  3. Perception
  4. Robotics
• Conclusions
  1. Philosophical Foundations
  2. AI: Present and Future

Some chapters are notably better than others. The game-playing chapter, for example, is excellent. It deals with search tree within board games - topics such as the various search algorithms, minimax, alpha-beta pruning, even expectimax were covered well. Other chapters of note are the introduction (provides a great overview of how AI came about from ancient philosophy to modern day linguistics and computer science), the neural/belief networks chapter, and many of the more specific agent chapters.

There is no source code anywhere in the book, making it an excellent choice for people who want a more language-neutral textbook. This isn't to say source code isn't available for the book - on the contrary, the authors provide Lisp and Python code themselves, and there are various ports to languages such as C++, Java and Prolog are readily available on the AIMA website.

One thing that did disappoint me was the lack of a chapter on evolutionary systems. They are, without a doubt, a very important section of modern Artificial Intelligence. Granted, since the book focuses more on agents, GAs and other evolutionary paradigms aren't immediately applicable, but it would have completed an otherwise fabulous textbook.