-Added MLP

NN/Exercise3.py

@@ -4,7 +4,7 @@
 
 start_data = np.array([[1, 0, 1],
  [0, 1, 1],
- [1, 1, 0],
+ [1, 1, 1],
  [0, 0, 0]])
 training_set = []
 
@@ -14,8 +14,6 @@
 print(training_set)
 
 def run_single_perceptron():
- f = open("weights.txt", "w")
-
  w = np.array([random.random(), random.random(), random.random()])
 
  learning_rate = 0.2
@@ -26,10 +24,12 @@ def calc_weights(weights, input, output, correct_output):
  return weights + (0.5*learning_rate*(correct_output-output)*input)
 
  epochs = 1000
- error = math.inf
- while error > 0.0001:
- for k in range(epochs):
+ success = False
+ avg_error = math.inf
+ for k in range(epochs):
+ if avg_error > 0.0001:
  np.random.shuffle(training_set)
+ 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])
@@ -41,10 +41,18 @@ def calc_weights(weights, input, output, correct_output):
  t = training_set[i][2]
  #error = 0.5*((t-v)**2) # squared error
  error = abs(t-v)
+ avg_error += error
  w = calc_weights(w, u, v, t)
-
- f.write(np.array_str(w))
- f.close()
+ avg_error = avg_error/len(training_set)
+ print(avg_error)
+ else:
+ success = True
+ print("Error threshold reached")
+ break
+ if success:
+ f = open("single_weights.txt", "w")
+ f.write(np.array_str(w))
+ f.close()
 
 
 def run_mlp():
@@ -99,7 +107,7 @@ def calc_weights(weights, error, input):
  print("Reached error threshold")
  break
  if success:
- f = open("weights.txt", "w")
+ f = open("mlp_weights.txt", "w")
  f.write(np.array_str(w_hidden[0]))
  f.write("\n")
  f.write(np.array_str(w_hidden[1]))
@@ -109,7 +117,7 @@ def calc_weights(weights, error, input):
 
 
 def testweights_mlp():
- f = open("weights.txt", "r")
+ f = open("mlp_weights.txt", "r")
  lines = f.readlines()
  f.close()
  w_hidden = []
@@ -140,15 +148,15 @@ def testweights_mlp():
 
 
 def testweights_single_perceptron(linenum):
- f = open("weights.txt", "r")
+ f = open("single_weights.txt", "r")
  lines = f.readlines()
  f.close()
  line = lines[linenum]
  line = line.replace("[", "")
  line = line.replace("]", "")
  w = np.fromstring(line, dtype=float, sep=" ")
  np.random.shuffle(training_set)
- for i in range(4):
+ 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)))
@@ -162,4 +170,4 @@ def testweights_single_perceptron(linenum):
 run_mlp()
 #testweights_mlp()
 #testweights_single_perceptron(0)
-#run()
+#run_single_perceptron()