Neural_Network_Library/Layer.cpp at master · EldonBird/Neural_Network_Library · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
#include <cmath>
#include <cstdlib>
#include <iostream>
#include <stdio.h>
#include <vector>


class Layer {

public:

    double* biases;
    double** weights;
    int* size_in;
    int* size_out;

    double** cost_gradient_w;
    double* cost_gradient_b;

    double* last_inputs;
    double* last_activations;
    double* last_weighted_inputs;

    Layer(int* si, int* so) {

        size_in = si;
        size_out = so;

        biases = new double[*size_out];
        last_activations = new double[*size_out];
        last_weighted_inputs = new double[*size_out];
        last_inputs = new double[*size_in];
        cost_gradient_b = new double[*size_out];

        weights = new double*[*size_in];
        for (int i = 0; i < *size_in; i++){ weights[i] = new double[*size_out]; }

        cost_gradient_w = new double*[*size_in];
        for (int i = 0; i < *size_in; i++){ cost_gradient_w[i] = new double[*size_out]; }

        start_with_random_weights();

    }
    Layer() {

    }


    ~Layer(){

        delete[] biases;
        delete[] cost_gradient_b;
        delete[] last_activations;
        delete[] last_weighted_inputs;
        delete[] last_inputs;
        for (int i = 0; i < *size_out; i++) { delete[] weights[i]; } delete[] weights;
        for (int i = 0; i < *size_out; i++) { delete[] cost_gradient_w[i]; } delete[] cost_gradient_w;
        delete size_in;
        delete size_out;

    }

    void start_with_random_weights() {
        for (int x = 0; x < *size_in; x++) {
            for (int y = 0; y < *size_out; y++) {
                double rng = ((double) rand() / RAND_MAX) * 2.0 - 1.0;
                weights[x][y] = rng / sqrt(*size_in);
            }
        }

        for (int i = 0; i < *size_out; i++) {
            biases[i] = ((double) rand() / RAND_MAX) * 0.1 - 0.05;
        }
    }

    void apply_gradients(double learn_rate) {

        for (int i = 0; i < *size_out; i++) {
            biases[i] -= cost_gradient_b[i] * learn_rate;
            for (int j = 0; j < *size_in; j++) {
                weights[i][j] -= cost_gradient_w[i][j] * learn_rate;
            }
        }
    }

    double* calculate_outputs(double* inputs) {

        for (int i = 0; i < *size_in; i++) {
            last_inputs[i] = inputs[i];
        }
        for (int x = 0; x < *size_out; x++) {
            double weighted_input = biases[x];
            for (int y = 0; y < *size_in; y++) {
                weighted_input += inputs[y] * weights[x][y];
            }
            last_weighted_inputs[x] = weighted_input;
            last_activations[x] = activation_function(weighted_input);
        }

        return last_activations;
    }

    double activation_function(double weighted_input) {
        return 1.0 / (1.0 + exp(-weighted_input));
    }

    double activation_function_derivative(double weighted_input) {

        double sigmoid = activation_function(weighted_input);
        return sigmoid * (1.0 - sigmoid);

    }

    double node_cost(double output_activation, double expected_output) {
        double error = output_activation - expected_output;
        return error * error;
    }

    double* hidden_layer_backpropigation(double* next_layer_error) {
        double* error = new double[*size_in];
        for (int i = 0; i < *size_in; i++ ) {
            error[i] = 0;
            for (int j =0; j < *size_out; j++) {
                error[i] += weights[i][j] * next_layer_error[j];
            }
        }

        return error;

    }

    void update_gradient(double* errors) {

        for (int i = 0; i < *size_out; i++) {
            double node_error = errors[i] * activation_function_derivative(last_weighted_inputs[i]);

            cost_gradient_b[i] = node_error;
            for (int j = 0; j < *size_in; j++) {
                cost_gradient_w[j][i] = last_inputs[j] * node_error;
            }
        }


    }

    double* calculate_output_layer(double* expected_outputs) {
        double* errors = new double[*size_in];

        for (int i = 0; i < *size_out; i++) {
            double cost_d = 2.0 * (last_activations[i] - expected_outputs[i]);
            errors[i] = cost_d * activation_function_derivative(last_weighted_inputs[i]);
        }

        return errors;
    }

};