Deletion of last_d_bias and last_d_weights

src/cnn/creation.c

@@ -29,7 +29,9 @@ Network* create_network(int max_size, int learning_rate, int dropout, int initia
     network->kernel[0]->nn = NULL; 
     network->kernel[0]->cnn = NULL; 
     create_a_cube_input_layer(network, 0, input_depth, input_dim); 
-    create_a_cube_input_z_layer(network, 0, input_depth, input_dim); 
+    // create_a_cube_input_z_layer(network, 0, input_depth, input_dim); 
+    // This shouldn't be used (if I'm not mistaken) so to save space, we can do:
+    ntework->input_z[0] = NULL; // As we don't backpropagate the input
     return network;
 }
 
@@ -104,7 +106,7 @@ void add_2d_average_pooling(Network* network, int dim_output) {
     network->kernel[k_pos]->nn = NULL;
     network->kernel[k_pos]->activation = 100*kernel_size; // Ne contient pas de fonction d'activation
     create_a_cube_input_layer(network, n, network->depth[n-1], network->width[n-1]/2);
-    create_a_cube_input_z_layer(network, n, network->depth[n-1], network->width[n-1]/2);
+    create_a_cube_input_z_layer(network, n, network->depth[n-1], network->width[n-1]/2); // Will it be used ?
     network->size++;
 }
 
@@ -130,33 +132,26 @@ void add_convolution(Network* network, int depth_output, int dim_output, int act
     cnn->columns = depth_output;
     cnn->w = (float****)malloc(sizeof(float***)*depth_input);
     cnn->d_w = (float****)malloc(sizeof(float***)*depth_input);
-    cnn->last_d_w = (float****)malloc(sizeof(float***)*depth_input);
     for (int i=0; i < depth_input; i++) {
         cnn->w[i] = (float***)malloc(sizeof(float**)*depth_output);
         cnn->d_w[i] = (float***)malloc(sizeof(float**)*depth_output);
-        cnn->last_d_w[i] = (float***)malloc(sizeof(float**)*depth_output);
         for (int j=0; j < depth_output; j++) {
             cnn->w[i][j] = (float**)malloc(sizeof(float*)*kernel_size);
             cnn->d_w[i][j] = (float**)malloc(sizeof(float*)*kernel_size);
-            cnn->last_d_w[i][j] = (float**)malloc(sizeof(float*)*kernel_size);
             for (int k=0; k < kernel_size; k++) {
                 cnn->w[i][j][k] = (float*)malloc(sizeof(float)*kernel_size);
                 cnn->d_w[i][j][k] = (float*)malloc(sizeof(float)*kernel_size);
-                cnn->last_d_w[i][j][k] = (float*)malloc(sizeof(float)*kernel_size);
             }
         }
     }
     cnn->bias = (float***)malloc(sizeof(float**)*depth_output);
     cnn->d_bias = (float***)malloc(sizeof(float**)*depth_output);
-    cnn->last_d_bias = (float***)malloc(sizeof(float**)*depth_output);
     for (int i=0; i < depth_output; i++) {
         cnn->bias[i] = (float**)malloc(sizeof(float*)*bias_size);
         cnn->d_bias[i] = (float**)malloc(sizeof(float*)*bias_size);
-        cnn->last_d_bias[i] = (float**)malloc(sizeof(float*)*bias_size);
         for (int j=0; j < bias_size; j++) {
             cnn->bias[i][j] = (float*)malloc(sizeof(float)*bias_size);
             cnn->d_bias[i][j] = (float*)malloc(sizeof(float)*bias_size);
-            cnn->last_d_bias[i][j] = (float*)malloc(sizeof(float)*bias_size);
         }
     }
     create_a_cube_input_layer(network, n, depth_output, bias_size);
@@ -188,14 +183,11 @@ void add_dense(Network* network, int output_units, int activation) {
     nn->output_units = output_units;
     nn->bias = (float*)malloc(sizeof(float)*output_units);
     nn->d_bias = (float*)malloc(sizeof(float)*output_units);
-    nn->last_d_bias = (float*)malloc(sizeof(float)*output_units);
     nn->weights = (float**)malloc(sizeof(float*)*input_units);
     nn->d_weights = (float**)malloc(sizeof(float*)*input_units);
-    nn->last_d_weights = (float**)malloc(sizeof(float*)*input_units);
     for (int i=0; i < input_units; i++) {
         nn->weights[i] = (float*)malloc(sizeof(float)*output_units);
         nn->d_weights[i] = (float*)malloc(sizeof(float)*output_units);
-        nn->last_d_weights[i] = (float*)malloc(sizeof(float)*output_units);
     }
     create_a_line_input_layer(network, n, output_units);
     create_a_line_input_z_layer(network, n, output_units);
@@ -227,14 +219,11 @@ void add_dense_linearisation(Network* network, int output_units, int activation)
     
     nn->bias = (float*)malloc(sizeof(float)*output_units);
     nn->d_bias = (float*)malloc(sizeof(float)*output_units);
-    nn->last_d_bias = (float*)malloc(sizeof(float)*output_units);
     nn->weights = (float**)malloc(sizeof(float*)*input_units);
     nn->d_weights = (float**)malloc(sizeof(float*)*input_units);
-    nn->last_d_weights = (float**)malloc(sizeof(float*)*input_units);
     for (int i=0; i < input_units; i++) {
         nn->weights[i] = (float*)malloc(sizeof(float)*output_units);
         nn->d_weights[i] = (float*)malloc(sizeof(float)*output_units);
-        nn->last_d_weights[i] = (float*)malloc(sizeof(float)*output_units);
     }
     /* Not currently used
     initialisation_1d_matrix(network->initialisation, nn->bias, output_units, output_units+input_units);

src/cnn/free.c

@@ -40,34 +40,27 @@ void free_convolution(Network* network, int pos) {
         for (int j=0; j < bias_size; j++) {
             free(k_pos->bias[i][j]);
             free(k_pos->d_bias[i][j]);
-            free(k_pos->last_d_bias[i][j]);
         }
         free(k_pos->bias[i]);
         free(k_pos->d_bias[i]);
-        free(k_pos->last_d_bias[i]);
     }
     free(k_pos->bias);
     free(k_pos->d_bias);
-    free(k_pos->last_d_bias);
 
     for (int i=0; i < r; i++) {
         for (int j=0; j < c; j++) {
             for (int k=0; k < k_size; k++) {
                 free(k_pos->w[i][j][k]);
                 free(k_pos->d_w[i][j][k]);
-                free(k_pos->last_d_w[i][j][k]);
             }
             free(k_pos->w[i][j]);
             free(k_pos->d_w[i][j]);
-            free(k_pos->last_d_w[i][j]);
         }
         free(k_pos->w[i]);
         free(k_pos->d_w[i]);
-        free(k_pos->last_d_w[i]);
     }
     free(k_pos->w);
     free(k_pos->d_w);
-    free(k_pos->last_d_w);
 
     free(k_pos);
 }
@@ -79,15 +72,12 @@ void free_dense(Network* network, int pos) {
     for (int i=0; i < dim; i++) {
         free(k_pos->weights[i]);
         free(k_pos->d_weights[i]);
-        free(k_pos->last_d_weights[i]);
     }
     free(k_pos->weights);
     free(k_pos->d_weights);
-    free(k_pos->last_d_weights);
 
     free(k_pos->bias);
     free(k_pos->d_bias);
-    free(k_pos->last_d_bias);
 
     free(k_pos);
 }
@@ -99,15 +89,12 @@ void free_dense_linearisation(Network* network, int pos) {
     for (int i=0; i < dim; i++) {
         free(k_pos->weights[i]);
         free(k_pos->d_weights[i]);
-        free(k_pos->last_d_weights[i]);
     }
     free(k_pos->weights);
     free(k_pos->d_weights);
-    free(k_pos->last_d_weights);
 
     free(k_pos->bias);
     free(k_pos->d_bias);
-    free(k_pos->last_d_bias);
 
     free(k_pos);
 }

src/cnn/include/function.h

@@ -48,6 +48,9 @@ void choose_apply_function_matrix(int activation, float*** input, int depth, int
 */
 void choose_apply_function_vector(int activation, float*** input, int dim);
 
+/*
+* Renvoie la fonction d'activation correspondant à son identifiant (activation)
+*/
 ptr get_function_activation(int activation);
 
 #endif

src/cnn/include/struct.h

@@ -7,10 +7,8 @@ typedef struct Kernel_cnn {
     int columns; // Depth of the output
     float*** bias; // bias[columns][dim_output][dim_output]
     float*** d_bias; // d_bias[columns][dim_output][dim_output]
-    float*** last_d_bias; // last_d_bias[columns][dim_output][dim_output]
     float**** w; // w[rows][columns][k_size][k_size]
     float**** d_w; // d_w[rows][columns][k_size][k_size]
-    float**** last_d_w; // last_d_w[rows][columns][k_size][k_size]
 } Kernel_cnn;
 
 typedef struct Kernel_nn {
@@ -18,10 +16,8 @@ typedef struct Kernel_nn {
     int output_units; // Nombre d'éléments en sortie
     float* bias; // bias[output_units]
     float* d_bias; // d_bias[output_units]
-    float* last_d_bias; // last_d_bias[output_units]
     float** weights; // weight[input_units][output_units]
     float** d_weights; // d_weights[input_units][output_units]
-    float** last_d_weights; // last_d_weights[input_units][output_units]
 } Kernel_nn;
 
 typedef struct Kernel {

test/cnn_convolution.cu

@@ -104,16 +104,13 @@ void run_convolution_test(int input_dim, int output_dim, int rows, int columns)
     // bias[kernel->columns][dim_output][dim_output]
     kernel->bias = create_matrix(kernel->columns, output_dim, output_dim, 15.0f);
     kernel->d_bias = create_matrix(kernel->columns, output_dim, output_dim, 1.5f);
-    kernel->last_d_bias = create_matrix(kernel->columns, output_dim, output_dim, 0.1f);
 
     // w[rows][columns][k_size][k_size]
     kernel->w = (float****)malloc(sizeof(float***)*kernel->rows);
     kernel->d_w = (float****)malloc(sizeof(float***)*kernel->rows);
-    kernel->last_d_w = (float****)malloc(sizeof(float***)*kernel->rows);
     for (int i=0; i < kernel->rows; i++) {
         kernel->w[i] = create_matrix(kernel->columns, kernel->k_size, kernel->k_size, 15.0f);
         kernel->d_w[i] = create_matrix(kernel->columns, kernel->k_size, kernel->k_size, 1.5f);
-        kernel->last_d_w[i] = create_matrix(kernel->columns, kernel->k_size, kernel->k_size, 0.1f);
     }
 
     float*** input = create_matrix(kernel->rows, input_dim, input_dim, 5.0f);
@@ -151,16 +148,13 @@ void run_convolution_test(int input_dim, int output_dim, int rows, int columns)
 
     free_matrix(kernel->bias, kernel->columns, output_dim);
     free_matrix(kernel->d_bias, kernel->columns, output_dim);
-    free_matrix(kernel->last_d_bias, kernel->columns, output_dim);
 
     for (int i=0; i < kernel->rows; i++) {
         free_matrix(kernel->w[i], kernel->columns, kernel->k_size);
         free_matrix(kernel->d_w[i], kernel->columns, kernel->k_size);
-        free_matrix(kernel->last_d_w[i], kernel->columns, kernel->k_size);
     }
     free(kernel->w);
     free(kernel->d_w);
-    free(kernel->last_d_w);
 
     free_matrix(input, kernel->rows, input_dim);
     free_matrix(output_cpu, kernel->columns, output_dim);