log_loss_op.cc 4.4 KB
Newer Older
K
kavyasrinet 已提交
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
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

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. */

#include "paddle/operators/log_loss_op.h"

namespace paddle {
namespace operators {

class LogLossOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

  void InferShape(framework::InferShapeContext* ctx) const override {
    PADDLE_ENFORCE(ctx->HasInput("Predicted"),
                   "Input(Predicted) must be initialized.");
    PADDLE_ENFORCE(ctx->HasInput("Labels"),
                   "Input(Labels) must be initialized.");

    auto pred_dims = ctx->GetInputDim("Predicted");
    auto label_dims = ctx->GetInputDim("Labels");

    PADDLE_ENFORCE_EQ(pred_dims, label_dims);
    PADDLE_ENFORCE_EQ(pred_dims.size(), 2,
                      "The rank of Input(Predicted) must be 2 and the shape is "
                      "[batch_size, 1].");
    PADDLE_ENFORCE_EQ(pred_dims[1], 1,
                      "Each row of Input(Predicted) contains a real value, "
                      "so the 2nd dimension of Input(X) must be 1.");

    ctx->SetOutputDim("Loss", {pred_dims[0], 1});
    ctx->ShareLoD("Predicted", "Loss");
  }
};

template <typename AttrType>
class LogLossOpMaker : public framework::OpProtoAndCheckerMaker {
 public:
  LogLossOpMaker(framework::OpProto* proto,
                 framework::OpAttrChecker* op_checker)
      : OpProtoAndCheckerMaker(proto, op_checker) {
    AddInput("Predicted",
             "The input value (Predicted) of Log loss op."
             "Predicted is a 2-D tensor with shape [batch_size, 1].");
    AddInput("Labels",
             "The target value (Labels) of Log loss op."
             "Labels is a 2-D tensor with shape [batch_size, 1].");
    AddOutput("Loss",
              "The output tensor with shape [batch_size, 1] "
              "which represents the log loss.");
    AddAttr<AttrType>("epsilon", "Epsilon in log loss.");
    AddComment(R"DOC(
LogLoss Operator.

Log loss is a loss function used for binary classification. Log Loss quantifies
the accuracy of a classifier by penalising false classifications. Minimising the
Log Loss is equivalent to maximising the accuracy of the classifier. We define
Predicted as the values predicted by our model and Labels as the target ground
truth value. Log loss can evaluate how close the predicted values are to the
target. The shapes of Predicted and Labels are both [batch_size, 1].
The equation is:

$$
Loss = - Labels * log(Predicted + \epsilon) -
        (1 - Labels) * log(1 - Predicted + \epsilon)
$$

)DOC");
  }
};

class LogLossGradOp : public framework::OperatorWithKernel {
 public:
  using framework::OperatorWithKernel::OperatorWithKernel;

  void InferShape(framework::InferShapeContext* ctx) const override {
    PADDLE_ENFORCE(ctx->HasInput("Predicted"),
                   "Input(Predicted) should not be null.");
    PADDLE_ENFORCE(ctx->HasInput("Labels"),
                   "Input(Labels) should not be null.");
    PADDLE_ENFORCE(ctx->HasInput(framework::GradVarName("Loss")),
                   "Input(Loss@GRAD) should not be null.");
    PADDLE_ENFORCE(ctx->HasOutput(framework::GradVarName("Predicted")),
                   "Output(Predicted@GRAD) should not be null.");

    auto pred_dims = ctx->GetInputDim("Predicted");
    auto label_dims = ctx->GetInputDim("Labels");
    auto loss_grad_dims = ctx->GetInputDim(framework::GradVarName("Loss"));
    PADDLE_ENFORCE_EQ(loss_grad_dims, pred_dims);

    auto pred_grad_name = framework::GradVarName("Predicted");
    ctx->SetOutputDim(pred_grad_name, pred_dims);
  }
};

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
REGISTER_OP(log_loss, ops::LogLossOp, ops::LogLossOpMaker<float>, log_loss_grad,
            ops::LogLossGradOp);
REGISTER_OP_CPU_KERNEL(
Q
QI JUN 已提交
113 114 115 116
    log_loss, ops::LogLossKernel<paddle::platform::CPUDeviceContext, float>);
REGISTER_OP_CPU_KERNEL(
    log_loss_grad,
    ops::LogLossGradKernel<paddle::platform::CPUDeviceContext, float>);