You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
94 lines
3.4 KiB
94 lines
3.4 KiB
import numpy as np
|
|
|
|
class AIlib:
|
|
def sigmoid(x):
|
|
return 1/(1 + np.exp(-x))
|
|
|
|
def correctFunc(inp:np.array): # generates the correct answer for the AI
|
|
return np.array( [inp[2], inp[1], inp[0]] ) # basically invert the rgb values
|
|
|
|
def calcCost( inp:np.array, out:np.array ): # cost function, lower -> good, higher -> bad, bad bot, bad
|
|
sumC = 0
|
|
outLen = len(out)
|
|
|
|
correctOut = AIlib.correctFunc(inp) # the "correct" output
|
|
|
|
for i in range(outLen):
|
|
sumC += (out[i] - correctOut[i])**2 # get the difference of every value
|
|
|
|
return sumC / outLen # return the cost
|
|
|
|
def genRandomMatrix( x:int, y:int, min: float=0.0, max: float=1.0 ): # generate a matrix with x, y dimensions with random values from min-max in it
|
|
# apply ranger with * and -
|
|
mat = np.random.rand(x, y) - 0.25
|
|
return mat
|
|
|
|
def think( inp:np.array, weights:list, bias:list, layerIndex: int=0 ): # recursive thinking, hehe
|
|
try:
|
|
maxLayer = len(weights) - 1
|
|
weightedInput = np.dot( inp, weights[layerIndex] ) # dot multiply the input and the weights
|
|
layer = AIlib.sigmoid( np.add(weightedInput, bias[layerIndex]) ) # add the biases
|
|
|
|
if( layerIndex < maxLayer ):
|
|
print(weights[layerIndex])
|
|
print("\n")
|
|
print("Layer " + str(layerIndex))
|
|
print(layer)
|
|
print("\n")
|
|
|
|
if( layerIndex < maxLayer ):
|
|
return AIlib.think( layer, weights, bias, layerIndex + 1 )
|
|
else:
|
|
out = np.squeeze(np.asarray(layer))
|
|
print("-Result-")
|
|
print(out)
|
|
print("\n")
|
|
return out
|
|
|
|
except (ValueError, IndexError) as err:
|
|
print("\n---------")
|
|
print( "Error: " + str(err) )
|
|
print( "Layer index: " + str(layerIndex) )
|
|
print( "Max layer index: " + str(maxLayer) )
|
|
|
|
def gradient( dCost:float, prop:list ):
|
|
propLen = len(prop)
|
|
gradient = [None] * propLen
|
|
for i in range( propLen - 1, -1, -1 ):
|
|
# if( i == propLen - 1 ):
|
|
# gradient[i] = dCost / prop[i]
|
|
# else:
|
|
# gradient[i] = dCost / (prop[i] + gradient[i+1])
|
|
gradient[i] = dCost / prop[i]
|
|
|
|
return gradient
|
|
|
|
def mutateProp( prop:list, gradient:list ):
|
|
newProp = [None] * len(gradient)
|
|
|
|
for i in range(len(gradient)):
|
|
newProp[i] = prop[i] - gradient[i] # * theta (relative to slope or something)
|
|
|
|
return newProp
|
|
|
|
def learn( inp:np.array, obj, theta:float ):
|
|
# Calculate the derivative for:
|
|
# Cost in respect to weights
|
|
# Cost in respect to biases
|
|
|
|
res1 = AIlib.think( inp, obj.weights, obj.bias ) # Think the first result
|
|
cost1 = AIlib.calcCost( inp, res1 ) # Calculate the cost of the thought result
|
|
|
|
inp2 = np.asarray( inp + theta ) # make the new input with `theta` as diff
|
|
res2 = AIlib.think( inp2, obj.weights, obj.bias ) # Think the second result
|
|
cost2 = AIlib.calcCost( inp2, res2 ) # Calculate the cost
|
|
|
|
dCost = cost2 - cost1 # get the difference
|
|
|
|
weightDer = AIlib.gradient( dCost, obj.weights )
|
|
biasDer = AIlib.gradient( dCost, obj.bias )
|
|
|
|
obj.weights = AIlib.mutateProp( obj.weights, weightDer )
|
|
obj.bias = AIlib.mutateProp( obj.bias, biasDer )
|
|
|
|
print("Cost: ", cost1)
|
|
|