Some cleanup

NN/Exercise3.py

@@ -4,7 +4,7 @@
 
 start_data = np.array([[1, 0, 1],
  [0, 1, 1],
- [1, 1, 1],
+ [1, 1, 0],
  [0, 0, 0]])
 training_set = []
 
@@ -18,9 +18,7 @@ def run_single_perceptron():
 
  learning_rate = 0.2
 
-
  def calc_weights(weights, input, output, correct_output):
- #return weights + learning_rate*output*(1- output)*(correct_output-output)*input # sigmoid
  return weights + (0.5*learning_rate*(correct_output-output)*input)
 
  epochs = 1000
@@ -32,8 +30,7 @@ def calc_weights(weights, input, output, correct_output):
  avg_error = 0
  for i in range(len(training_set)):
  u = np.append(training_set[i][:2], -1)
- z = (w[0]*u[0]) + (w[1]*u[1]) + (w[2]*u[2])
- #v = 1 / (1+(math.e**(-z)))
+ z = np.matmul(w, u)
  if z <= 0:
  v = 0
  else:
@@ -82,8 +79,8 @@ def calc_weights(weights, error, input):
  u = np.append(training_set[i][:2], -1)
  x_1 = w_hidden[0]
  x_2 = w_hidden[1]
- z_1 = (x_1[0] * u[0]) + (x_1[1] * u[1]) + (x_1[2] * u[2])
- z_2 = (x_2[0] * u[0]) + (x_2[1] * u[1]) + (x_2[2] * u[2])
+ z_1 = np.matmul(x_1, u)
+ z_2 = np.matmul(x_2, u)
  x_1_output = (1 / (1+(math.e**(-z_1))))
  x_2_output = (1 / (1+(math.e**(-z_2))))
  z = (w_output[0] * x_1_output) + (w_output[1] * x_2_output) + (w_output[2] * -1)
@@ -135,8 +132,8 @@ def testweights_mlp():
  u = np.append(training_set[i][:2], -1)
  x_1 = w_hidden[0]
  x_2 = w_hidden[1]
- z_1 = (x_1[0] * u[0]) + (x_1[1] * u[1]) + (x_1[2] * u[2])
- z_2 = (x_2[0] * u[0]) + (x_2[1] * u[1]) + (x_2[2] * u[2])
+ z_1 = np.matmul(x_1, u)
+ z_2 = np.matmul(x_2, u)
  x_1_output = 1 / (1 + (math.e ** (-z_1)))
  x_2_output = 1 / (1 + (math.e ** (-z_2)))
  z = (w_output[0] * x_1_output) + (w_output[1] * x_2_output) + (w_output[2] * -1)
@@ -158,16 +155,15 @@ def testweights_single_perceptron(linenum):
  np.random.shuffle(training_set)
  for i in range(10):
  u = np.append(training_set[i][:2], -1)
- z = (w[0] * u[0]) + (w[1] * u[1]) + (w[2] * u[2])
- #v = 1 / (1 + (math.e ** (-z)))
+ z = np.matmul(w, u)
  if z <= 0:
  v = 0
  else:
  v = 1
  print("Input:", u, " Output:", v)
 
 
-run_mlp()
+#run_mlp()
 #testweights_mlp()
 #testweights_single_perceptron(0)
 #run_single_perceptron()