tiny_dnn  1.0.0
A header only, dependency-free deep learning framework in C++11
arithmetic_layer.h
1 /*
2  Copyright (c) 2013, Taiga Nomi
3  All rights reserved.
4 
5  Redistribution and use in source and binary forms, with or without
6  modification, are permitted provided that the following conditions are met:
7  * Redistributions of source code must retain the above copyright
8  notice, this list of conditions and the following disclaimer.
9  * Redistributions in binary form must reproduce the above copyright
10  notice, this list of conditions and the following disclaimer in the
11  documentation and/or other materials provided with the distribution.
12  * Neither the name of the <organization> nor the
13  names of its contributors may be used to endorse or promote products
14  derived from this software without specific prior written permission.
15 
16  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
17  EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
18  WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
19  DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY
20  DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
21  (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
22  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
23  ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24  (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
25  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27 #pragma once
28 #include "tiny_dnn/util/util.h"
29 #include "tiny_dnn/layers/layer.h"
30 
31 namespace tiny_dnn {
32 
36 class elementwise_add_layer : public layer {
37 public:
42  elementwise_add_layer(serial_size_t num_args, serial_size_t dim)
43  : layer(std::vector<vector_type>(num_args, vector_type::data), {vector_type::data}), num_args_(num_args), dim_(dim) {}
44 
45  std::string layer_type() const override {
46  return "elementwise-add";
47  }
48 
49  std::vector<shape3d> in_shape() const override {
50  return std::vector<shape3d>(num_args_, shape3d(dim_,1,1));
51  }
52 
53  std::vector<shape3d> out_shape() const override {
54  return{ shape3d(dim_,1,1) };
55  }
56 
57  void forward_propagation(const std::vector<tensor_t*>& in_data,
58  std::vector<tensor_t*>& out_data) override {
59  const tensor_t& in1 = *in_data[0];
60  tensor_t& out = *out_data[0];
61 
62  out = in1;
63 
64  // @todo parallelize
65  for (size_t sample = 0; sample < in1.size(); ++sample) {
66  for (serial_size_t i = 1; i < num_args_; i++) {
67  std::transform((*in_data[i])[sample].begin(),
68  (*in_data[i])[sample].end(),
69  out[sample].begin(),
70  out[sample].begin(),
71  [](float_t x, float_t y){ return x + y; });
72  }
73  }
74  }
75 
76  void back_propagation(const std::vector<tensor_t*>& in_data,
77  const std::vector<tensor_t*>& out_data,
78  std::vector<tensor_t*>& out_grad,
79  std::vector<tensor_t*>& in_grad) override {
80  CNN_UNREFERENCED_PARAMETER(in_data);
81  CNN_UNREFERENCED_PARAMETER(out_data);
82  for (serial_size_t i = 0; i < num_args_; i++)
83  *in_grad[i] = *out_grad[0];
84  }
85 
86  template <class Archive>
87  static void load_and_construct(Archive & ar, cereal::construct<elementwise_add_layer> & construct) {
88  serial_size_t num_args, dim;
89 
90  ar(cereal::make_nvp("num_args", num_args), cereal::make_nvp("dim", dim));
91  construct(num_args, dim);
92  }
93 
94  template <class Archive>
95  void serialize(Archive & ar) {
96  layer::serialize_prolog(ar);
97  ar(cereal::make_nvp("num_args", num_args_), cereal::make_nvp("dim", dim_));
98  }
99 private:
100  serial_size_t num_args_;
101  serial_size_t dim_;
102 };
103 
104 } // namespace tiny_dnn
105 
element-wise add N vectors y_i = x0_i + x1_i + ...
Definition: arithmetic_layer.h:36
std::vector< shape3d > in_shape() const override
array of input shapes (width x height x depth)
Definition: arithmetic_layer.h:49
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
return delta of previous layer (delta=\frac{dE}{da}, a=wx in fully-connected layer)
Definition: arithmetic_layer.h:76
std::vector< shape3d > out_shape() const override
array of output shapes (width x height x depth)
Definition: arithmetic_layer.h:53
elementwise_add_layer(serial_size_t num_args, serial_size_t dim)
Definition: arithmetic_layer.h:42
std::string layer_type() const override
name of layer, should be unique for each concrete class
Definition: arithmetic_layer.h:45
void forward_propagation(const std::vector< tensor_t * > &in_data, std::vector< tensor_t * > &out_data) override
Definition: arithmetic_layer.h:57
base class of all kind of NN layers
Definition: layer.h:62