New Book Now in Editing
For the past few months I've been working on a new book. Actually, its the second edition to one of my most popular books. Introduction to Neural Networks has been out since 2005, and its time for a second edition.
The first edition focused on using the JOONE engine, and unfortunatly the JOONE engine has changed API formats several times over to break my book's examples. The second edition focuses on showing YOU how to create a neural network in Java, without using an off-the-shelf network. You will be shown, step-by-step how to construct and train several types of neural networks. JOONE is still covered, but the current version of JOONE is confined to one chapter. The new book also includes about 4 times the number of examples of the first book. Which was always a weakness of the first edition. Additionally new topics such as neural network prediction, financial markets, and neural networks with bots are covered. This really is a fairly large-scale rewrite of the original text.
As of yesterday I am done with the book's examples and neural network code and most of the book text. Editing has begun. I don't like to give exact deadlines on books. But I am planning on a release sometime this summer. Once the Java book is out, I plan to release a second book, on the same topic. Introduction to Neural Networks with C#. I will probably use Mono as my base compatability level. Mono is a very exciting technology that allows C# to be developed on non-Microsoft platforms. I would see keeping the book's source code Mono compatable as a plus.
- JeffHeaton's blog
- Login or register to post comments

C# Version
The Java version is about 1/2 edited at this point. Probably September that it will be out. C# version will be out likely by the end of the year. Maybe sooner.
I am currently using C# 2008 to write the C# version. Most examples are console, so I expect them to work with Mono okay. I need to learn more about Mono, but my understanding is that the GUI side of it is not complete. So my goal for Mono compatability will be that the core neural network code works in it. I will post more about this as I get the code working.
I am also really trying to have the code look like real C# code and not Java code that was just translated to C#. Namely, all the getters/setters replaced by properties, using indexers and delegates where they make sense. There are a number of features to C# that Java just does not have and I am making sure we use them where it makes sense.
Question
"I will probably use Mono as my base compatability level."
Does this mean that I can use C# as usual not worrying about Mono or does it mean that I need to use Mono? I guess my question gave me away as a newbie but I am a newbie with ambitions...or dreams. I guess I could learn Java but I would love to be able to stick to C# which is hard enough for an economist. What time frame is the C# version looking at.