View article

The authors meant this book to be used as a classroom text or for self-study of topics in knowledge acquisition (also known as machine learning). The first half of the book fulfills that purpose moderately well. The algorithms based on statistical methods and the well-known programs that embody them, such as CART and ID3, are described at some length in chapters 1 through 4. Chapter 5 gives a much more abbreviated look at topics like explanation-based learning and genetic algorithms. Chapter 6 briefly covers some of what is known about human learning. Since how humans learn is too poorly understood to provide much guidance for machine learning, this chapter could have been omitted.(Besides, there is no reason why a computer system should learn the way humans do anyway). The remainder of the book, nearly half, discusses the authors' work on rule extraction from neural nets. However interesting this …