I learned on Ted Serbinski that Scott Reynolds, who implements the Slope One Collaborative Filtering algorithm in Drupal, is being paid by Google. It is one of 14 projects that Google decided to fund. Way to go Google!

Click on the picture to see it all. Don’t ask me what Louka ate, I have no idea. You can see Lohan is quite aware of the camera. My wife is enjoying a good book.

Update: According to my wife, Louka was eating sand. Gross!

Olivier Bousquet questions Statistical Learning Theory.

So, I think that the theoretical analysis is mainly a way to popularize an algorithm and to raise its visibility. The effect is then that more people try it out, and streamline it. So in the end, the algorithm may be adopted, but a theoretical analysis rarely justifies the algorithm and never provides guarantees. Hence theory is a way to attract attention to an algorithm. It should also be a way to get new insights for the development of new algorithms, but this happens much less frequently than is claimed!

I’m currently working on a paper where I spent a lot of time fiddling and proving some bounds, as Olivier points out we often do. Of course, not being a pure theory guy, I also run extensive experiments. I cannot help but notice a strong disconnect between the theory and the practice, not only in what I do, but in the papers I cite.

The counter-argument might be: but theory helps design better algorithms. Like Olivier, I don’t buy it.

Ah! I still do theory of course, but let’s not fool ourselves in thinking we are smarter than nature.

Scott Reynolds has been working hard to implement Slope One collaborative filtering in Drupal. You can see his latest code in viewcvs.

The Springer Online Mathematical Encyclopedia is really very interesting. The quality is amazing.

The Online Encyclopaedia of Mathematics is the most up-to-date and comprehensive English-language graduate-level reference work in the field of mathematics today. This online edition comprises more than 8,000 entries and illuminates nearly 50,000 notions in mathematics. The Encyclopaedia of Mathematics is updated on a regular basis to remain a quick, precise source of reference to mathematical definitions, concepts, explanations, surveys, examples, terminology and methods, which will prove useful for all mathematicians and other scientists who encounter mathematics in their work.

Just read Doron Zeilberger’s 72nd Opinion. The man is quite a bit pretentious (”mathematician is really another species, higher than homo sapiens”) though he doesn’t lose one bit of respect from me.

Back in 2002, I had a discussion, in a café in the old town of Saint Malo, where I basically tried to convey the following message, but not as well put:

(…) most of human mathematics is completely useless. It was developed by humans for human consumption. In order for humans to understand it, it had to proceed in tiny steps, each comprehensible to a human. But if we take the “mesh size” of each step, dA, to be larger, one can do potentially much bigger and better things, and the computer’s dA is much larger, so we can (potentially) reach a mountain-top much faster, and conquer new mountain-tops where no humans will ever tread with their naked brains.

So this activity of computer-generated mathematics is the future. Unfortunately, many human mathematicians still don’t realize the importance of this activity, and dismiss it as “just a computer program” and “no new mathematics”.

At the time, the mathematician I was talking to, a man I greatly respect, objected that nobody could predict what could be useful, so to claim that non-computable math. is worth less than computable math., was just foolish. It managed to silence me. Indeed, while I believe that algorithms are a higher form of mathematics, I cannot prove that it is, and neither can Zeilberger, but he makes a great case for it:

[computable math] is a methodology that will make all computer-free math obsolete very soon.

I didn’t see this pass by, but ACM has launched several new journals. At least some of these will be top tier. Here are some titles I like:

• Transactions on Knowledge Discovery in Data
  
• Transactions on the Web
  
• Transactions on Storage
  

This can only be a good thing.