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
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
|
/*
Copyright (c) 2013, Taiga Nomi
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the <organization> nor the
names of its contributors may be used to endorse or promote products
derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY
DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#pragma once
#include "tiny_dnn/util/util.h"
#include "tiny_dnn/layers/layer.h"
#include <algorithm>
namespace tiny_dnn {
/**
* applies dropout to the input
**/
class dropout_layer : public layer {
public:
typedef activation::identity Activation;
typedef layer Base;
/**
* @param in_dim [in] number of elements of the input
* @param dropout_rate [in] (0-1) fraction of the input units to be dropped
* @param phase [in] initial state of the dropout
**/
dropout_layer(serial_size_t in_dim, float_t dropout_rate, net_phase phase = net_phase::train)
: Base({vector_type::data}, {vector_type::data}),
phase_(phase),
dropout_rate_(dropout_rate),
scale_(float_t(1) / (float_t(1) - dropout_rate_)),
in_size_(in_dim)
{
mask_.resize(1, std::vector<uint8_t>(in_dim));
clear_mask();
}
dropout_layer(const dropout_layer& obj) = default;
virtual ~dropout_layer(){}
#ifdef CNN_USE_DEFAULT_MOVE_CONSTRUCTORS
dropout_layer(dropout_layer&& obj) = default;
dropout_layer& operator=(const dropout_layer& obj) = default;
dropout_layer& operator=(dropout_layer&& obj) = default;
#endif
void set_dropout_rate(float_t rate)
{
dropout_rate_ = rate;
scale_ = float_t(1) / (float_t(1) - dropout_rate_);
}
float_t dropout_rate() const {
return dropout_rate_;
}
///< number of incoming connections for each output unit
serial_size_t fan_in_size() const override
{
return 1;
}
///< number of outgoing connections for each input unit
serial_size_t fan_out_size() const override
{
return 1;
}
std::vector<index3d<serial_size_t>> in_shape() const override {
return{ index3d<serial_size_t>(in_size_, 1, 1) };
}
std::vector<index3d<serial_size_t>> out_shape() const override {
return{ index3d<serial_size_t>(in_size_, 1, 1) };
}
void back_propagation(const std::vector<tensor_t*>& in_data,
const std::vector<tensor_t*>& out_data,
std::vector<tensor_t*>& out_grad,
std::vector<tensor_t*>& in_grad) override {
tensor_t& prev_delta = *in_grad[0];
const tensor_t& curr_delta = *out_grad[0];
CNN_UNREFERENCED_PARAMETER(in_data);
CNN_UNREFERENCED_PARAMETER(out_data);
for (serial_size_t sample = 0; sample < static_cast<serial_size_t>(prev_delta.size()); ++sample) {
for (serial_size_t i = 0; i < static_cast<serial_size_t>(curr_delta.size()); i++) {
prev_delta[sample][i] = mask_[sample][i] * curr_delta[sample][i];
}
}
}
void forward_propagation(const std::vector<tensor_t*>& in_data,
std::vector<tensor_t*>& out_data) override {
const tensor_t& in = *in_data[0];
tensor_t& out = *out_data[0];
const size_t sample_count = in.size();
if (mask_.size() < sample_count) {
mask_.resize(sample_count, mask_[0]);
}
for (size_t sample = 0, sample_count = in.size(); sample < sample_count; ++sample) {
std::vector<uint8_t>& mask = mask_[sample];
const vec_t& in_vec = in[sample];
vec_t& out_vec = out[sample];
if (phase_ == net_phase::train) {
for (size_t i = 0; i < in_vec.size(); i++)
mask[i] = bernoulli(dropout_rate_);
for (size_t i = 0; i < in_vec.size(); i++)
out_vec[i] = mask[i] * scale_ * in_vec[i];
}
else {
for (size_t i = 0, end = in_vec.size(); i < end; i++)
out_vec[i] = in_vec[i];
}
}
}
/**
* set dropout-context (training-phase or test-phase)
**/
void set_context(net_phase ctx) override
{
phase_ = ctx;
}
std::string layer_type() const override { return "dropout"; }
// currently used by tests only
const std::vector<uint8_t>& get_mask(serial_size_t sample_index) const {
return mask_[sample_index];
}
void clear_mask() {
for (auto& sample : mask_) {
std::fill(sample.begin(), sample.end(), 0);
}
}
template <class Archive>
static void load_and_construct(Archive & ar, cereal::construct<dropout_layer> & construct) {
net_phase phase;
float_t dropout_rate;
serial_size_t in_size;
ar(cereal::make_nvp("in_size", in_size), cereal::make_nvp("dropout_rate", dropout_rate), cereal::make_nvp("phase", phase));
construct(in_size, dropout_rate, phase);
}
template <class Archive>
void serialize(Archive & ar) {
layer::serialize_prolog(ar);
ar(cereal::make_nvp("in_size", in_size_), cereal::make_nvp("dropout_rate", dropout_rate_), cereal::make_nvp("phase", phase_));
}
private:
net_phase phase_;
float_t dropout_rate_;
float_t scale_;
serial_size_t in_size_;
std::vector<std::vector<uint8_t>> mask_;
};
} // namespace tiny_dnn
|
