Merge remote-tracking branch 'origin/master'

Exercise2.py

@@ -16,7 +16,7 @@
  [' ', '#', ' ', ' ', ' '],
  [' ', '#', ' ', ' ', ' ']]
 
-environment = complex_environment
+environment = simple_environment
 
 #Dimensions of the environment
 COLUMNS = len(environment[0])
@@ -34,7 +34,7 @@
 action_selected_matrix = [[-1 for x in range(COLUMNS)] for y in range(ROWS)]
 step_size = 0.1 # Alpha
 discount_rate = 0.9 # Gamma
-epsilon = 0.1 # ϵ for greedy
+epsilon = 1 # ϵ for greedy
 episode_amount = 100
 reward_for_each_episode = [0 for x in range(episode_amount)]
 index = 0

NN/Perceptron.m

@@ -1,9 +1,25 @@
-T = now;
-rng(T)
+
 
 bias = 0;
-u = [bias, 0,0; bias,0,1; bias,1,0; bias,1,1]
+u = [bias, 0,0; bias,0,1; bias,1,0; bias,1,1];
 v = 0;
 t = [0, 0, 0, 1];
-weigths = -1 + (1+1)*rand(1,2)
-w = [-1, weigths]
+weigths = -1 + (1+1)*rand(1,2);
+w = [-1, weigths];
+error = inf;
+epochs = 1000;
+
+while error > 0.001
+ u
+ for epoch = 1:epochs
+ matrix_input_output = [u,t'];
+ Shuffle(matrix_input_output);
+ matrix_input_output
+ t = matrix_input_output(:,end)';
+ u = matrix_input_output(:, 1:end-1);
+ for i = 1:4
+ input_vector = u(i,:);
+ correct_output = t(i);
+ end
+ end
+end

NN/Shuffle.m

@@ -0,0 +1,5 @@
+ function v=shuffle(v)
+ rng shuffle
+ v=v(randperm(length(v)));
+ end
+