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
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
|
// Copyright 2016 The TensorFlow Authors. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// =============================================================================
// TreePredictions returns the per-class probabilities for each input by
// evaluating the given tree.
#include <algorithm>
#include "tensorflow/contrib/tensor_forest/kernels/data_spec.h"
#include "tensorflow/contrib/tensor_forest/kernels/tree_utils.h"
#include "tensorflow/core/framework/op.h"
#include "tensorflow/core/framework/op_kernel.h"
#include "tensorflow/core/framework/shape_inference.h"
#include "tensorflow/core/kernels/bounds_check.h"
#include "tensorflow/core/util/work_sharder.h"
namespace tensorflow {
using tensorforest::CHILDREN_INDEX;
using tensorforest::FEATURE_INDEX;
using tensorforest::LEAF_NODE;
using tensorforest::FREE_NODE;
using tensorforest::CheckTensorBounds;
namespace {
// Traverse the tree for every example from start to end. Put the resulting
// prediction probability into output_predictions[i].
void Evaluate(OpKernelContext* context,
const std::function<bool(int, int32, float)>& decide,
const Tensor& weights, const Tensor& tree_tensor,
const Tensor& tree_thresholds, int valid_leaf_threshold,
Tensor* output_predictions, int32 start, int32 end) {
auto out = output_predictions->tensor<float, 2>();
const auto node_pcw = weights.tensor<float, 2>();
const auto tree = tree_tensor.tensor<int32, 2>();
const auto thresholds = tree_thresholds.unaligned_flat<float>();
const int32 num_classes = static_cast<int32>(weights.shape().dim_size(1));
const int32 num_nodes = static_cast<int32>(tree_tensor.shape().dim_size(0));
for (int i = start; i < end; i++) {
int node_index = 0;
int parent = -1;
while (true) {
OP_REQUIRES(context, FastBoundsCheck(node_index, num_nodes),
errors::InvalidArgument("node_index not in valid range."))
const int32 left_child = tree(node_index, CHILDREN_INDEX);
if (left_child == LEAF_NODE) {
const int32 flat_leaf_index = node_index * num_classes + 1;
const int32 flat_parent_index = parent * num_classes + 1;
std::vector<float> means(num_classes - 1);
tensorforest::GetParentWeightedMean(
node_pcw(node_index, 0), node_pcw.data() + flat_leaf_index,
node_pcw(parent, 0), node_pcw.data() + flat_parent_index,
valid_leaf_threshold, num_classes - 1, &means);
const int32 start_index = i * (num_classes - 1);
std::copy(means.begin(), means.end(), out.data() + start_index);
break;
} else if (left_child == FREE_NODE) {
LOG(ERROR) << "Reached a free node, not good.";
return;
}
parent = node_index;
const int32 feature = tree(node_index, FEATURE_INDEX);
node_index = left_child + decide(i, feature, thresholds(node_index));
}
}
}
} // namespace
class TreePredictions : public OpKernel {
public:
explicit TreePredictions(OpKernelConstruction* context)
: OpKernel(context) {
OP_REQUIRES_OK(context, context->GetAttr(
"valid_leaf_threshold", &valid_leaf_threshold_));
string serialized_proto;
OP_REQUIRES_OK(context, context->GetAttr("input_spec", &serialized_proto));
input_spec_.ParseFromString(serialized_proto);
}
void Compute(OpKernelContext* context) override {
const Tensor& input_data = context->input(0);
const Tensor& sparse_input_indices = context->input(1);
const Tensor& sparse_input_values = context->input(2);
const Tensor& sparse_input_shape = context->input(3);
const Tensor& tree_tensor = context->input(4);
const Tensor& tree_thresholds = context->input(5);
const Tensor& node_per_class_weights = context->input(6);
int32 num_data = 0;
if (sparse_input_indices.shape().dims() == 2) {
const auto sparse_shape = sparse_input_shape.unaligned_flat<int64>();
// TODO(gilberth): This is because we can't figure out the shape
// of a sparse tensor at graph-build time, even if the dimension is
// actually known.
input_spec_.mutable_sparse(0)->set_size(sparse_shape(1));
num_data = sparse_shape(0);
OP_REQUIRES(context, sparse_input_values.shape().dims() == 1,
errors::InvalidArgument(
"sparse_input_values should be one-dimensional"));
OP_REQUIRES(context, sparse_input_shape.shape().dims() == 1,
errors::InvalidArgument(
"sparse_input_shape should be one-dimensional"));
OP_REQUIRES(context,
sparse_input_indices.shape().dim_size(0) ==
sparse_input_values.shape().dim_size(0),
errors::InvalidArgument(
"sparse_input_indices and sparse_input_values should "
"agree on the number of non-zero values"));
OP_REQUIRES(context,
sparse_input_indices.shape().dim_size(1) ==
sparse_input_shape.shape().dim_size(0),
errors::InvalidArgument(
"sparse_input_indices and sparse_input_shape should "
"agree on the dimensionality of data points"));
}
if (input_data.shape().dim_size(0) > 0) {
const int32 dense_num_data =
static_cast<int32>(input_data.shape().dim_size(0));
if (num_data > 0) {
CHECK_EQ(num_data, dense_num_data)
<< "number of examples must match for sparse + dense input.";
}
num_data = dense_num_data;
OP_REQUIRES(
context, input_data.shape().dims() == 2,
errors::InvalidArgument("input_data should be two-dimensional"));
}
OP_REQUIRES(context, tree_tensor.shape().dims() == 2,
errors::InvalidArgument(
"tree should be two-dimensional"));
OP_REQUIRES(context, tree_thresholds.shape().dims() == 1,
errors::InvalidArgument(
"tree_threhsolds should be one-dimensional"));
OP_REQUIRES(context, node_per_class_weights.shape().dims() == 2,
errors::InvalidArgument(
"node_pcw should be two-dimensional"));
OP_REQUIRES(
context,
tree_tensor.shape().dim_size(0) ==
tree_thresholds.shape().dim_size(0) &&
tree_tensor.shape().dim_size(0) ==
node_per_class_weights.shape().dim_size(0),
errors::InvalidArgument(
"Number of nodes should be the same in "
"tree, tree_thresholds and node_pcw."));
// Check tensor bounds.
if (!CheckTensorBounds(context, input_data)) return;
if (!CheckTensorBounds(context, sparse_input_indices)) return;
if (!CheckTensorBounds(context, sparse_input_values)) return;
if (!CheckTensorBounds(context, sparse_input_shape)) return;
if (!CheckTensorBounds(context, tree_tensor)) return;
if (!CheckTensorBounds(context, tree_thresholds)) return;
if (!CheckTensorBounds(context, node_per_class_weights)) return;
const int32 num_classes = static_cast<int32>(
node_per_class_weights.shape().dim_size(1));
Tensor* output_predictions = nullptr;
TensorShape output_shape;
output_shape.AddDim(num_data);
output_shape.AddDim(num_classes - 1);
OP_REQUIRES_OK(context,
context->allocate_output(0, output_shape,
&output_predictions));
// Lambdas to capture the eigen-tensors so we don't the conversion overhead
// on each call to DecideNode.
const auto get_dense = tensorforest::GetDenseFunctor(input_data);
const auto get_sparse = tensorforest::GetSparseFunctor(sparse_input_indices,
sparse_input_values);
auto decide = [&get_dense, &get_sparse, this](int example, int32 feature,
float threshold) {
return tensorforest::DecideNode(get_dense, get_sparse, example, feature,
threshold, input_spec_);
};
auto worker_threads = context->device()->tensorflow_cpu_worker_threads();
int num_threads = worker_threads->num_threads;
const int64 costPerUnit = 800;
auto work = [context, &decide, &node_per_class_weights, &tree_tensor,
&tree_thresholds, this, &output_predictions,
num_data](int64 start, int64 end) {
CHECK(start <= end);
CHECK(end <= num_data);
Evaluate(context, decide, node_per_class_weights, tree_tensor,
tree_thresholds, valid_leaf_threshold_, output_predictions,
static_cast<int32>(start), static_cast<int32>(end));
};
Shard(num_threads, worker_threads->workers, num_data, costPerUnit, work);
}
private:
float valid_leaf_threshold_;
tensorforest::TensorForestDataSpec input_spec_;
};
REGISTER_KERNEL_BUILDER(Name("TreePredictions").Device(DEVICE_CPU),
TreePredictions);
} // namespace tensorflow
|
