Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
From curve fitting to machine learning : an illustrative guide to scientific data analysis and computational intelligence (2nd ed.)
Zielesny A., Springer International Publishing, New York, NY, 2016. 498 pp. Type: Book (978-3-319325-44-6)
Date Reviewed: Feb 8 2017

In an age where machine learning (ML) and its many manifestations are among the hottest buzzwords around, a title like this would certainly appeal to anyone. In a broad sense, ML is some kind of curve fitting, particularly when the system is driven primarily by observations, so the title was appealing.

The book consists of five main chapters. The first chapter, “Introduction,” provides an overall motivation, and covers topics such as optimization, model function, and structure of data (and challenges in making it usable) for various algorithms. These are useful to know if you are working in ML. Chapter 2, “Curve Fitting,” discusses the various issues in identifying the right kind of curve to fit a set of points. The initial parts of this chapter are done well, taking readers step by step through getting the right fit for your curve. Chapter 3, “Clustering,” discusses various clustering approaches. ML is the topic of chapter 4, which covers a subset of the popular ML algorithms. ML is seen as identifying an approximate function for mapping given inputs to desired outputs. Techniques like neural networks, support vector machines (SVM), and regression are covered. This chapter occupies almost one-third of the book. Chapter 5, “Discussion,” is on various crosscutting concerns like data smoothing, the scope of machine learning, and so on. The appendix discusses a computational intelligence package (CIP), which is used throughout the book.

I have several issues with this book. First, despite the title, there is almost nothing in the book about computational intelligence. Even the term “data analysis” in the subtitle is overkill. The book has a much narrower focus. Providing code segments corresponding to major functionalities or algorithms is useful for the reader who wants to try out variations, and for making the idea discussed more concrete. But the book takes this to an extreme, and is full of code and its output plots. Many details and variations are brought out in the same format; most of these are best left to the reader to experiment, with proper pointers. Since the plots are not explained, one can get lost as to what look for. Since most of the technicalities are left in the code and the plot, the dependency on CIP is very high. Even a version change of the package can make the book unusable. There is no theory covered for most of the topics, making the material hard to relate to.

The ML chapter has a skewed orientation, without discussing the general principles and approaches. Not all ML can be usefully viewed as finding a mathematical function. Models like inductive learning and version space models are not mentioned.

Even for CIP, the coverage in the appendix is more as a reference; no introduction is provided. None of the major topics or approaches has any introduction in the text, nor is any attempt made to connect the various topics. There are no exercises. Finally, the index contains many common words, such as “problem,” “number,” and “options.”

Overall, though it has some nice aspects, I found this book difficult to use. It is more of a doer’s book for those who know CIP and want to dabble with aspects of ML in CIP. Given that there are plenty of available books in areas such as data science, analytics, and ML, this is a book you can do without.

Reviewer:  M Sasikumar Review #: CR145048 (1705-0259)
Bookmark and Share
  Reviewer Selected
Featured Reviewer
 
 
Learning (I.2.6 )
 
 
Applications And Expert Systems (I.2.1 )
 
 
Physical Sciences And Engineering (J.2 )
 
Would you recommend this review?
yes
no
Other reviews under "Learning": Date
Learning in parallel networks: simulating learning in a probabilistic system
Hinton G. (ed) BYTE 10(4): 265-273, 1985. Type: Article
Nov 1 1985
Macro-operators: a weak method for learning
Korf R. Artificial Intelligence 26(1): 35-77, 1985. Type: Article
Feb 1 1986
Inferring (mal) rules from pupils’ protocols
Sleeman D.  Progress in artificial intelligence (, Orsay, France,391985. Type: Proceedings
Dec 1 1985
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy