Am I late to WPF - Windows Presentation Foundation? I think I am, but not for long. Some pretty amazing applications are built with in WPF especially Mindmapping applications. There is even a Brisbane based company, NovaMind building cool Mindmapping software for both Mac and Windows Vista. Gives me hope that cool jobs do exist in Brisbane! I look for something more interesting than database driven development! So here’s my list of things that I’d like to do in my spare time to make 2008 more interesting:

• Data mining, machine learning, social network formation and analysis
• Firefox extensions
• Diagramming applications
  Heres where WPF will come in handy! I think this will be fun and a good break from the database driven application boredom!
• Roleplay Edu apps
  WWF - Windows Workflow foundation would make a superb foundation for building roleplay applications (really just workflow but not for business). It even integrates perfectly with ASP.NET. I’ll be posting on this again soon.

So if you are looking for a freelancer to do projects that fit into any of the broad categories I outlined above, drop me a line…..

There was a time when all I spoke about was SVMs! That was a fun time and perhaps I should have continued my PHD and not gotten consumed by my day job (where I built a tool to translate content to the Blackboard XML import/export). Will I ever get over it????? Maybe if I complain more!!!

SVM’s are now easier than ever to use without having to implement the complex Mathematics yourself. Jeez — the Ruby and Python libs put SVMs at your fingertips. Check out this article on using SVM and Ruby to implement a SPAM filter. SVMs are great at complex classification eg document classification. This could be the most enlightening thing you read all year and the year is still young. Save yourself from the CRUD that is web development. Save yourself while you can!!!

Over the last few months, I’ve noticed something quite incredible - machine learning techniques slowly (but surely) creeping into mainstream conferences. It all started with the Machine Learning Made Easy with Perl presentation at OSCON 2007 by Lino Ramirez. Then the amazing Programming Collective Intelligence book was published. And now straight from RailsConf 2007 comes Making Rails More (Artificially) Intelligent which is quite filled with Ruby code richness. The presentation covers Bayesian Classifiers and Genetic Algorithms. A GA was used to “optimize web page revenues from advertisements” as an example.

Its interesting to note that all three examples that I quote use a dynamic language - Perl, Python and Ruby.

….now where are those AI in Web 2.0 style jobs in Brisbane, Australia? Where?????

I declared that “Programming Collective Intelligence” was the must have book of 2007 even before I received my copy. I’ve spent the last 2 weeks reading and working through the examples. I am pleased to say that it has lived up to all expectations. The explanations are clear, the examples are relevant and the code is easily re-usable. I’ve tried to read numerous other Machine Learning books and papers only to be confused by the mathematics. “Programming Collective Intelligence” provides a great introduction to Machine Learning in the context of creating social networking application. I’d recommend the book for anybody new to Machine Learning or with a general interest in making their applications a little bit smarter. Machine Learning is a vast and complex field, so be warned that to fully understand the algorithms and implement the latest advances, you’ll need to visit more advanced books and academic papers (Maths, Maths and More Maths).

Project Highlights:

• Building a del.icio.us Link Recommender
• Clustering and Filtering Blog posts based on word usage
• Using Decision Trees to predicting signups
• Use Support Vector Machines for matchmaking
  (…. never thought of using SVM on a dating site … )
• Simplicity of Python libraries used
  (Including the Imaging Library used to draw a Dendrogram and the wrapper for LIBSVM)

Go and buy the book right now!

Programming Collective Intelligence is my “must have” book for 2007. I’ve been looking for a book that was able to explain machine learning, data mining and the related mathematics. I think I’ve found it! This book seems to cover just about everything including:

• Recommendations (Collaborative and Item based filtering)
• Supervised and Unsupervised Learning
• Clustering
• Searching and Ranking
• Decision Trees
• Kernel Methods and Support Vector Machines (SVM)
• Genetic Programming

I hope I finally understand the maths behind SVMs. I’m off to order the book. I’ll post a review soon…..