kilitary / fann-related Goto Github PK

using ta-lib (for example) and genetic algorithm (for improved selection when best results reached), reach optimized network with less time and work.
for example such program will try all functions from math library to combine it in different ways.
and using genetic algorithm for parameter choosing, when success reached.

1. select ta-lib outputs (need to write parser of .h) and dynamically calling its all possible functions to find algorithm wich the train shows as best adopted for this train data. looks like reverse approach.
   maybe someone created this already, but noone knows.
   like choosing random auto parts when you did not know wich is best, create the algo that will try it ALL in different logic blocks, successfully combine and test drived in train on train and new data.
2. for example one indicator kicks another and program see that MSE/HIT is changed, then using genetic algorithm wilth frequently choosed best ranked solutions (Functions)
   with the highest rank.
3. the genetic algorithm i mention is that was in MT4/MT5 tester (forex client),