Algorithms Books

Wonderful well thought out book, I keep in in office as a reference. Easy layout and comprehensive. Good enough for the professional, but well written enough for the lay person.

Taking a longer view that some other cryptography books, this book still covers all the main issues in current cryptography. But what the author is trying to do is give a broader overview, that is not tied to specific cryptosystems that are currently deployed.

All the good stuff is here - elliptic methods, el Gamal, hashing/hash digests, Public Key Infrastructure etc.

Perhaps this book can be compared to Bruce Schneier's groundbreaking "Applied Cryptography" of over ten years ago. Both books deliberately refrained from tying their textual descriptions too closely to contemporary implementations.

The mathematical treatment in this book is perhaps a little deeper than Schneier's book, but still far easier for most users than a fully fledged math treatise.

The most interesting parts may be Rhee's speculations on the future directions of cryptography. Potentially, if you are a researcher, this may be the most useful to you, with potential indicators of where to go.

For some readers, the best attraction of the book is the GUI program that lets you quickly experiment and apply the main ideas. The text is very interdisciplinary, written for diverse audiences, spanning biology, computer science and mathematics. Some aspects of the book may perhaps be too mathematical for some biology readers. Say the Hidden Markov Models, for example. But if you keep at it, you should get able to get the gist of the models. Which is another reason for the usefulness of the GUI. Essentially, so long as you understand the basic math ideas, the GUI lets you sidestep the grotty details and focus on applying the overall models.

It could also be that the book is suitable for a university course. The chapter exercises and accompanying answers are useful, in this regard.

The book is about abstractions and algorithms for distributed programming. Arguably, the abstractions are considered to be at least as important as the algorithms. The abstractions effectively are assumptions that let you push down certain details into lower level code. A very modular approach that allows for a solid object oriented design.

Notably, you should not have to worry about the semantics of pushing bytes from one machine to another. Network programming is now so well established that you won't have to write code to move a byte set. So there is a stable, debugged network protocol stack. (Not necessarily TCP/IP, to be sure.)

The algorithms in the book can then focus on such matters as how to distribute data and collect results across a network of machines. Very importantly, to also detect when a machine might have timed out or otherwise generated a failure.

The book covers various optimization techniques in Engineering but not all.However,the analysis of each algorithm is great with flowcharts and mathematical equations accompanying each subject.
There are many books for Iterative computer algorithms but this one combines all the necessary information to help you develop and test your own algorithm.

This book covers "local" machine learning methods (k-NN, local regression, etc.) as a collection of distinct contributions. The material is current and useful.

Two drawbacks that kept me from giving 5 stars: too little comparison to alternative techniques (nerual networks, parametric regression, tree induction, etc.) and the lack of an index.

I've been using this book to teach introductory mathematical logic to computer science students for many years. Every year or two, I look again for something that might cover the ground better, but, despite some notational flaws and an occasional error or poor choice of phrase, this one comes out on top every time.

This very recent book can be used as a follow on the Davies tome, Machine Vision : Theory, Algorithms, Practicalities. The latter was a good general treatment of the many ideas used in image recognition. Whereas the current book does not start from scratch in the field, and takes the reader to much of the current research issues.

Image segmentation, morphology and edge extraction are several of these issues. The level of mathematical detail goes beyond the introductory style of Davies. Which can be a great help if you are relying on Davies. The problem with it is that a serious researcher often has to recreate or rederive much of the needed low level methods. These are often presented here by Steger et al.

For users of OpenCV, note that it is nowhere referenced in the book. But some ideas that form its basis, like template matching, can be found here.

For a reader already versed in the topics of this book, it presents a useful historical perspective. Giving an appreciation of what our predecessors fumbled through.

The first topic has perhaps the strongest historical tie-in. Relating problems that bedevilled classical European and Asian mathematicians to the rise of calculus. These concered the sums of infinite series. Pondering this problem with no definitive solution was the status quo for centuries. It took the genius of Euler to finally furnish an answer. For the modern reader, this narrative shows one of Euler's masterpieces of deduction.

Another topic is about numerical solutions of linear and non-linear equations. Today this is germane because of our computers. Yet, hard as it is to imagine, the methods were also pondered centuries ago, when all this had to be done by hand.

This book is a gem of problem sets AND solutions, in the field of algorithms. The problems were from actual examinations given at Stanford in various computer science classes. About half the book is good descriptive text about the ideas that the problems probe. Certainly, well written, as befits Knuth's contribution.

But I would suggest to you that the best use of the book is in tackling those problems. In the spirit of Knuth's classic Art of Computer Programming series, where he gives extensive questions and answers.

I realise my suggestion may have appeal to only some of you. But I'm addressing my remarks to the smartest amongst you glancing at this. Test and improve your understanding of algorithms.