From 6a6fe40730cbb76801a1250e4fc95fbb985d55ec Mon Sep 17 00:00:00 2001
From: Luis Marzoratti <lmarzora@itba.edu.ar>
Date: Tue, 9 May 2017 20:11:57 -0300
Subject: [PATCH] linear output

---
 multilayer_perceptron.m | 22 ++++++++++++++++------
 test.m                  | 12 ++++++------
 2 files changed, 22 insertions(+), 12 deletions(-)

diff --git a/multilayer_perceptron.m b/multilayer_perceptron.m
index 69ac288..bcece9a 100644
--- a/multilayer_perceptron.m
+++ b/multilayer_perceptron.m
@@ -28,16 +28,16 @@
     %setup
     for i = 2:length(neurons_per_layer)
         m = m + 1;
-        weights{m} = (rand(neurons_per_layer(i), neurons_per_layer(i-1)+1) .- 0.5)./neurons_per_layer(i-1);
+        %weights{m} = (2*(rand(neurons_per_layer(i), neurons_per_layer(i-1)+1) .- 0.5))./100;
+        weights{m} = (rand(neurons_per_layer(i), neurons_per_layer(i-1)+1) .- 0.5)./(neurons_per_layer(i-1));
         layer_entry{m} = [-1, zeros(1, neurons_per_layer(i-1))];
     	h{m} = [-1 ,zeros(1, neurons_per_layer(i-1))];
     end
-
     %last layer
     M = m;
 
     for iteration = 1:max_iterations
-        
+    tic; 
 	%select random entry
         for index = randperm(n);
             %get layers output 
@@ -51,9 +51,19 @@
                 fflush(1);
 	        end
             h{M} = weights{M} * layer_entry{M}';
-            output(index) = activation_func(h{M});
+            
+            %no linear
+            %output(index) = activation_func(h{M});
+            %get errors
+            %d{M} = activation_der(h{M})*(expected_output(index) - output(index));
+            
+            %linear
+            output(index) = h{M};
+            %h{M};
             %get errors
-            d{M} = activation_der(h{M})*(expected_output(index) - output(index));
+            d{M} = (expected_output(index) - output(index));
+            %d{M};
+            
             for i = M-1:-1:1
                 d{i} = (activation_der(h{i})' .* (d{i+1} * weights{i+1}(:,2:end)));
             end
@@ -66,7 +76,7 @@
         end
         %get iteration error
         error_per_iteration(iteration) = sum((expected_output - output).^2)/n;
-        error_per_iteration(iteration)
+        [error_per_iteration(iteration),iteration,toc]
         fflush(1);
         if error_per_iteration(iteration) <= tolerance
             return
diff --git a/test.m b/test.m
index 3bdeeb9..2223542 100644
--- a/test.m
+++ b/test.m
@@ -1,20 +1,20 @@
 1;
 source multilayer_perceptron.m
 terrain = dlmread('terrain/terrain03.data');
-[f,fder] = activation_exp(1);
-net = [2,45,50,1];
-[weights,output,mse] = multilayer_perceptron_learn(terrain(:,1:2)',terrain(:,3)',net ,f,fder,1,1e4,5e-4);
+[f,fder] = activation_tanh(1);
+net = [2,200,50,10,1];
+[weights,output,mse] = multilayer_perceptron_learn(terrain(:,1:2)',terrain(:,3)',net ,f,fder,.05,1e4,2e-4);
 x = [-3:0.01:3];
 y = x;
 %dlmwrite('weights.csv',cell2mat(weights));
 
 for i = 1:length(x)
         for j = 1:length(y)
-                out(i,j) = get_output([x(i);y(j)],weights,net,f);
+            z(i,j) = get_output([x(i);y(j)],weights,net,f); 
         end
 end
 
-plot3(terrain(:,1),terrain(:,2),terrain(:,3),'.')
+plot3(terrain(:,1),terrain(:,2),terrain(:,3),'.','markersize',12)
 hold on;
-surf(x,y,out);
+surf(x,y,z);
 hold off;