NeuralNetwork/Serialize.cpp at master · jbgrim/NeuralNetwork · 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
#include "Network.h"

#include <ostream>
#include <sstream>

namespace NeuralNetwork {

// Encoding/decoding functions
// =============================
vector<vector<vector<float>>> deserialize(istream& is)
{
	vector<vector<vector<float>>> layers;
	string line;
	getline(is, line, '\n');
	istringstream iss(line);
	string str_layer;
	while (getline(iss, str_layer, '*'))
	{
		istringstream iss_layer(str_layer);
		string str_neuron;
		vector<vector<float>> layer;
		while (getline(iss_layer, str_neuron, '/'))
		{
			istringstream iss_neuron(str_neuron);
			string weight;
			vector<float> neuron;
			while (getline(iss_neuron, weight, ' '))
			{
				neuron.push_back(stof(weight));
			}
			layer.push_back(neuron);
		}
		layers.push_back(layer);
	}
	return layers;
}

uint8_t to_int8(float f)
{
	f = f > 1.f ? 1.f : f;
	f = f < -1.f ? -1.f : f;
	return static_cast<int8_t>(f * 126);
}

uint8_t to_int8(const char f)
{
	switch (f)
	{
	case ' ':
		return 0b10000000;
	case '/':
		return 0b10000001;
	case '*':
		return 0b01111111;
	default:  // should not happen
		return 0;
	}
}

float to_float(const uint8_t i)
{
	return static_cast<float>(i) / 126.f;
}

vector<uint8_t> serialize(const vector<vector<vector<float>>>& data)
{
	vector<uint8_t> serialized;
	for (size_t i = 0; i < data.size(); ++i)
	{
		for (size_t j = 0; j < data[i].size(); ++j)
		{
			for (const float k : data[i][j])
			{
				serialized.push_back(to_int8(k));
			}
			if (j != data[i].size() - 1)
			{
				serialized.push_back(to_int8('/'));
			}
		}
		if (i != data.size() - 1)
		{
			serialized.push_back(to_int8('*'));
		}
	}
	return serialized;
}

vector<vector<vector<float>>> deserialize(const vector<uint8_t>& serialized)
{
	vector layers(1, vector(1, vector<float>()));
	int layer = 0;
	int neuron = 0;
	for (const unsigned char param : serialized)
	{
		if (param == to_int8('*'))
		{
			++layer;
			neuron = 0;
			layers.emplace_back(1, vector<float>());
		}
		else if (param == to_int8('/'))
		{
			++neuron;
			layers[layer].emplace_back();
		}
		else
		{
			layers[layer][neuron].push_back(to_float(param));
		}
	}
	return layers;
}

vector<bool> decode(int rec, const int n_layers)
{
	vector<bool> recs(n_layers);
	for (int i = 0; i < n_layers; ++i)
	{
		recs[i] = rec & 1;
		rec >>= 1;
	}
	return recs;
}

int encode(const vector<bool>& rec)
{
	int recs = 0;
	for (int i = static_cast<int>(rec.size()) - 1; i >= 0; --i)
	{
		recs <<= 1;
		recs += rec[i];
	}
	return recs;
}

// Printing function
ostream& Network::print_to(ostream& os) const
{
	const auto old_precision = os.precision();
	os.precision(3);
	for (size_t i = 0; i < m_layers_.size(); ++i)
	{
		os << "layer " << i << ":\n";
		for (auto& layer : m_layers_[i])
		{
			os << "\t";
			for (const float value : layer)
			{
				os << value << "\t";
			}
			os << "\n";
		}
	}
	os.precision(old_precision);
	return os;
}
}