Complex Pattern Finding i Large Data Sets

In the later part of my PhD work I found that a bayesian recurrent neural network (a network using Bayesian logic and a frequentist approach to set the weights) which was originally invented by Anders Lansner and Örjan Ekeberg in the SANS group and has been improved by, Anders Holst and Anders Sandberg is an excellent tool to perform data mining in large data sets. To find patterns tremendously quick, due to the fact that it is working on global statistics. A method which has been succesfully applied to find syndromes in the WHO database.

This is published in the thesis I will soon provide a link to here, but due to the copyright rules and that the paper is in the publishing process I am not allowed, according the rules of today, to provide the actual paper.

The invention which I will present at this site before April 23rd 2004 builds upon this method which can be seen as applying AI in Medicine (AIM). I have extended this with another type of rule based reasoning into a method which will be applied to two other areas (which may also be abbreviated AIM). Two areas were researchers and geek thinking people are frustrated today, but where this invention will help to cure the reason for these symptoms which makes us frustrated.

Problems: (almost similar to the early warning problem)

• Even if the method is great it is unfortunately not possible to publish on the web due to the copyright rules.
• In this kind of publications there is a tremendous time lag. The paper was originally submitted in Jan 2003, returned May 2003, resubmitted in Aug 2003, returned again for fixes in Feb 2003.
• When a paper is submitted, it is "locked in" until the reviers of the journal has told their view, in this process several years can easily pass by.
• The paper is submitted to a journal specializing in neural networks, but this is quite a narrow area so the paper will most likely not be read by most researchers and applications which could make use of this technique.
• Even though research results shold be free for all they are in practice locked in.
• It is not possible to store this kind of knowledge in a big data base and for instance perform data mining on, due to the "locking in". To purchase all papers that would be needed would cost a tremendous amount.
• Formats for this kind of information exchange are not standardized which makes it very hard to do anything meaningful of them. The formats which are used today are almost completely lacking structure. I use TeX, some use Word, but these are both almost useless for information retriveal viewpoint. Once SGML was used which allowed HTML to be created. HTML is just another unstructured mess, only usable for final presentation. HTML is not an information preserving format. XML is a subset of SGML but XML is less general than SGML. HTML is a document type in SGML but HTML can not be specified using XML for instance (XHTML can however due to the strict requirements of end tags in XHTML).
• Why, with improved technology, should we go backwards?
• If I write a similar paper and did not send it to the publisher, I could then publish it on the web, but how could you then know that what I've written wouldn't be crap, as it would not be peer reviewed.