Some changes

Joseph Redmon

2015-11-26 0305fb4d99cf1efc7d4aa4d2ee2d65d54500d437

 src/network.c

@@ -8,10 +8,10 @@

#include "crop_layer.h"
#include "connected_layer.h"
#include "local_layer.h"
#include "convolutional_layer.h"
#include "deconvolutional_layer.h"
#include "detection_layer.h"
#include "region_layer.h"
#include "normalization_layer.h"
#include "maxpool_layer.h"
#include "avgpool_layer.h"
@@ -26,18 +26,41 @@
    return batch_num;
}

void reset_momentum(network net)
{
    if (net.momentum == 0) return;
    net.learning_rate = 0;
    net.momentum = 0;
    net.decay = 0;
    #ifdef GPU
        if(gpu_index >= 0) update_network_gpu(net);
    #endif
}

float get_current_rate(network net)
{
    int batch_num = get_current_batch(net);
    int i;
    float rate;
    switch (net.policy) {
        case CONSTANT:
            return net.learning_rate;
        case STEP:
            return net.learning_rate * pow(net.gamma, batch_num/net.step);
            return net.learning_rate * pow(net.scale, batch_num/net.step);
        case STEPS:
            rate = net.learning_rate;
            for(i = 0; i < net.num_steps; ++i){
                if(net.steps[i] > batch_num) return rate;
                rate *= net.scales[i];
                if(net.steps[i] > batch_num - 1) reset_momentum(net);
            }
            return rate;
        case EXP:
            return net.learning_rate * pow(net.gamma, batch_num);
        case POLY:
            return net.learning_rate * pow(1 - (float)batch_num / net.max_batches, net.power);
        case SIG:
            return net.learning_rate * (1./(1.+exp(net.gamma*(batch_num - net.step))));
        default:
            fprintf(stderr, "Policy is weird!\n");
            return net.learning_rate;
@@ -49,6 +72,8 @@
    switch(a){
        case CONVOLUTIONAL:
            return "convolutional";
        case LOCAL:
            return "local";
        case DECONVOLUTIONAL:
            return "deconvolutional";
        case CONNECTED:
@@ -61,8 +86,6 @@
            return "softmax";
        case DETECTION:
            return "detection";
        case REGION:
            return "region";
        case DROPOUT:
            return "dropout";
        case CROP:
@@ -104,12 +127,12 @@
            forward_convolutional_layer(l, state);
        } else if(l.type == DECONVOLUTIONAL){
            forward_deconvolutional_layer(l, state);
        } else if(l.type == LOCAL){
            forward_local_layer(l, state);
        } else if(l.type == NORMALIZATION){
            forward_normalization_layer(l, state);
        } else if(l.type == DETECTION){
            forward_detection_layer(l, state);
        } else if(l.type == REGION){
            forward_region_layer(l, state);
        } else if(l.type == CONNECTED){
            forward_connected_layer(l, state);
        } else if(l.type == CROP){
@@ -144,6 +167,8 @@
            update_deconvolutional_layer(l, rate, net.momentum, net.decay);
        } else if(l.type == CONNECTED){
            update_connected_layer(l, update_batch, rate, net.momentum, net.decay);
        } else if(l.type == LOCAL){
            update_local_layer(l, update_batch, rate, net.momentum, net.decay);
        }
    }
}
@@ -169,10 +194,6 @@
            sum += net.layers[i].cost[0];
            ++count;
        }
        if(net.layers[i].type == REGION){
            sum += net.layers[i].cost[0];
            ++count;
        }
    }
    return sum/count;
}
@@ -213,12 +234,12 @@
            backward_dropout_layer(l, state);
        } else if(l.type == DETECTION){
            backward_detection_layer(l, state);
        } else if(l.type == REGION){
            backward_region_layer(l, state);
        } else if(l.type == SOFTMAX){
            if(i != 0) backward_softmax_layer(l, state);
        } else if(l.type == CONNECTED){
            backward_connected_layer(l, state);
        } else if(l.type == LOCAL){
            backward_local_layer(l, state);
        } else if(l.type == COST){
            backward_cost_layer(l, state);
        } else if(l.type == ROUTE){
@@ -319,6 +340,7 @@
    //if(w == net->w && h == net->h) return 0;
    net->w = w;
    net->h = h;
    int inputs = 0;
    //fprintf(stderr, "Resizing to %d x %d...", w, h);
    //fflush(stderr);
    for (i = 0; i < net->n; ++i){
@@ -332,9 +354,12 @@
            break;
        }else if(l.type == NORMALIZATION){
            resize_normalization_layer(&l, w, h);
        }else if(l.type == COST){
            resize_cost_layer(&l, inputs);
        }else{
            error("Cannot resize this type of layer");
        }
        inputs = l.outputs;
        net->layers[i] = l;
        w = l.out_w;
        h = l.out_h;
@@ -525,12 +550,12 @@
    return acc;
}

float *network_accuracies(network net, data d)
float *network_accuracies(network net, data d, int n)
{
    static float acc[2];
    matrix guess = network_predict_data(net, d);
    acc[0] = matrix_topk_accuracy(d.y, guess,1);
    acc[1] = matrix_topk_accuracy(d.y, guess,5);
    acc[0] = matrix_topk_accuracy(d.y, guess, 1);
    acc[1] = matrix_topk_accuracy(d.y, guess, n);
    free_matrix(guess);
    return acc;
}