Hi there! I have some problem with self organizing maps. I've got a lot of these patterns: www.maxis-house.de/files/m1.gif www.maxis-house.de/files/m2.gif www.maxis-house.de/files/m3.gif every picture shows 6 patterns. I now want to use SOMs to classify these patterns, because I want to know, which kinds of patterns exist. I think SOMs are really good for that and I want to train myself using SOMs :) So my question is: Which would be the best way to reveal the pattern data to a self-organizing map (or to any other neural network)? every pattern has that red line, and every pattern consists of the green lines, but not every pattern has the same number of lines. A line is represented by its starting and ending point, but it does not matter which point is the beginning and wich point is the end of the line. So how can I solve that problem?