Thread: Decoding Algorithm

I'm making a library (I haven't decided over *.lib or *.dll yet - I have experience with *.lib, but I'm told *.dll has advantages - I'm not sure if I need those advantages, and I have software suitable for both) consisting of a single function (one code - several functions to allow for different types). at function is passed as many dimensions of arrays of any type. Then I go through and using my Super Duper Specia Top Secret Algorithm, decode it by finding patterns by studying differences between each array's element and even between each array/dimension. I have a feeling I'm going to need some basic AI concepts, and the most AI I have ever done was a program in BASIC when I was 8 that used the RND function and a bunch of arrays to make PC generated art. Some of it looked pretty cool - but the concepts I used aren't very applicable to much else. I've checked AI Horizon's site, and it doesn't exactly have any concepts that fit what I'm looking for. Anyone have any suggestions that I can try beyond studying changes?

Well I figured it out - I'll use this: I call it the differential pyramid:
I just start out finding difference between each element and array, and eventually, it funnels down to a single number. That is always used as a factor in the equation for the pattern, and the number of levels is used as an exponent in the equation for th pattern. Then I have a neural network set up to try different solution using what I will have at this point, and to monitor what's working best. I'll have to modify the concept of the network I'm using, as I will have to prevent it from ever ruling out any solution entirely. Anyone got any suggestions for whether or not to use either library?