Swarm behavior, neural computation, photos, thoughts. And an ant colony in a terrarium.

As you might have recognized, there have been a few technical issues at this place – for instance, it had become pretty slow (up to 9 secs per request at peak times), and some bug in the old blog engine sometimes seriously messed up the home page layout. Now, there's help:

Software overhaul. I changed the blog engine to BlogTNG, which solved the home page messing up issue. Unfortunately, this changed the feed urls (however, the new ones are already embedded in the pages, so you'll be fine). There may be some issues with the new RSS streams during the first days, sorry for that.

New design and lay-out. Everything (in particular, the home page) is more slick and clean now. I almost completely rewrote the site design and also changed some parts of navigation and lay-out. See for yourselves

New Hardware. The site now got a dedicated dual-core server, so you are free to place load now . I hope we withstand even load peaks without major delays – this hasn't been the case so far.

The english version has been completely overhauled, the entire manuscript got a dedicated reference implementation (SNIPE), a brand-new resilient propagation section and much more. It's winter term (at least in Germany) and there is a new major release of my manuscript “A Brief Introduction to Neural Networks”. As usual, it is available for download here.

1. The English version has been thoroughly and completely overhauled. Almost every single sentence has been touched. Lots and lots of still present German werds have been removed, clumsy speech has been removed, paragraphs rewritten. Lots of thanks go to Sebastian Merzbach!
2. SNIPE has been integrated into the manuscript in form of an official reference implementation. In particular throughout chapters 3, 4 and 5 are lots of special shaded paragraphs who guide the reader how to implement their context using Snipe. Despite the easy handling of Snipe classes and methods due to their speaking names, Snipe is really fast and efficient due to its original goal of implementation. Therefore, I hope that the learning success of those readers who read my manuscript for later implementation of neural networks is increased significantly.
3. The text has been cleaned of unneccessary style elements, e.g. environments like “remark” or “example” have been removed. They were still a relict of the times when I tried out lots of things in latex. Of course, the contents of those environments are still there, however the unneccessary structural elements that have been there every few paragraphs are now gone.
4. Chapter 5: A new section deals with the learning procedure Resilient Propagation as an extensioon of Backpropagatioon fo Error. The arrangement of this chapter has been overworked as well.
5. Chapters 3, 4, 5: Definitions have been straightened, as well as the text has been.
6. The preface has been shortened, straightened and more clearly deals with how to read the manuscript in a senseful way.
7. The cover has been slightly renewed and now, there is a nice QR code.
8. Lots and lots of small errors have been removed right across the entire manuscript, just because readers sent me e-mails – among those also were technical errors. For example, some diacritical letters could not be found via full-text search, a bug which has been solved.

In particular to the last bullet point I have to state how encouraging your e-mails are. In the mean time, I have been getting e-mails from readers several times a week, providing me feed-back, correcting typos, asking interesting questions or just commendations.

I want to encourage everybody to continue writing me e-mails, because it keeps up my motivation in writing. Every little error you guys are mailing me will be added to my big manuscript todo list and will be implemented in the next release, if appropriate. And of course, you get an answer.

Thanks, and let's continue the good work!

In this version, the internal data structure of snipe got some flexible extendability features. What's more, the training procedure resilient propagation was overhauled ( this includes an interface change). You can download snipe as usual by going to the snipe page.

• Dynamic synapse shadows have been added. The data structure was generalized (without slowing it down, no worries). But after all, what is a dynamic synapse shadow? It's pretty easy. In the documentation of Snipe, I mentioned a specially designed, pretty fast and flexible data structure for the storage of synapse data. A shadow now offers an additional double data cell for every single synapse. Arbitrarily many of such shadows can be added. This is very useful for e.g. training procedures like backpropagation, that need to store a learning rate or something els per synapse. Now that we know what a shadow is, the word “dynamic” remains to explain. If the network gets changed in structure (e.g. by removing a neuron), the shadows will be adapted automatically. At the moment, I implemented several operators to deal with synapse shadows, without really documenting them. However, they will be the fundament for a clean and slick inheritance policy which may be introduced in the next release.
• Resilient Backpropagation was overhauled. Note: This update causes a change in interface concerning the method com.dkriesel.snipe.core.NeuralNetwork.trainResilientBackpropagation. A big thank you for the overhaul goes to Martin Westhoven. He not only managed to fix a bug that might have prevented resilient propagation from learning (no other components were influenced), but also implemented a further Rprop-enhancement that you can activate if you like (have a look at the documentation). Furthermore, Rprop has been adapted to the synapse shadow structure.
• Several classes are serializable now.

Have a lot of fun!

Nice microscope photos of very small living beings, the 38 best of 2000 images! Pretty nice to see. The whole BigPicture-Site is also nice. (Thx, Verena)

Last week, I have attended a conference in Paris. In the evenings, there was time for some photos, some of which I now post here.

Read more…

Axel Boldt, a German college tracher living in the US, compares his experiences of living in the US and Germany in a nicely readable way. He sheds light on positive and negative aspects to encounter in both of the countries. This is the text I would wish I had written after my US experience in 2007.

Our Institute moved again, and as a consequence I have a much larger office now. Thus, my old photo ensembles were not enough to fill the walls. This is why I added two ensembles.

Read more…

Don't try this at home: What happens if you put your head into the beam of a particle accelerator. Wow . TL;DR: This is a shorter article in Wikipedia..

As promised, I lightroomed some of the – in my opinion – most beautiful photos of New Zealand.

Read more…

We're back Hope I get the top photos chosen and the panoramas chosen soon!