def TrainAGeneration(BrainList, ScrambleList, NumToReturn):
BrainFitnesses = []
for i in range(0, len(BrainList)):
#print("Brain: "+str(i))
Fitness = 0
for j in range(0, len(ScrambleList)):
NewCube = Cube()
NewCube.PerformPredeterminedScramble(ScrambleList[j])
Fitness += AttemptSolve(NewCube, BrainList[i])
#print("BrainFitness: "+str(Fitness))
BrainFitnesses.append(Fitness)
BrainList[i].Fitness = Fitness
BrainsToReturn = []
for i in range(0, NumToReturn):
MaxFitness = max(BrainFitnesses)
BrainsToReturn.append(BrainList[BrainFitnesses.index(MaxFitness)])
BrainList.pop(BrainFitnesses.index(MaxFitness))
BrainFitnesses.pop(BrainFitnesses.index(MaxFitness))
return BrainsToReturn
from copy import deepcopy
import HelperFunctions
#training 50 brains on one cube.
if __name__ == "__main__":
#initialize 50 random brains
TheBrains = []
for i in range(0, 50):
TheBrains.append(Brain())
#try brains with 2 internal layers, each with 50 neurons
TheBrains[i].InitializeRandomLayerNeurons(2, 50)
#Determine the initial scrambles for our cubes
ScrambleList = []
for i in range(0, 1):
ScrambleList.append(HelperFunctions.GenerateScramble(20))
#test ten generations
for Generation in range(0, 59):
print()
print("Testing Generation: " + str(Generation))
BestBrains = HelperFunctions.TrainAGeneration(TheBrains, ScrambleList,
5)
TheBrains = []
for BrainElement in BestBrains:
TheBrains = TheBrains + HelperFunctions.MakeNewGenerationFromBrain(
BrainElement, 10)
BestBrains = HelperFunctions.TrainAGeneration(
TheBrains, ScrambleList, 1) #Here is our leading contender
DemoCube = Cube()
DemoCube.PerformPredeterminedScramble(ScrambleList[0])
HelperFunctions.PerformExhibitionSolve(DemoCube, BestBrains[0])
