paddle中新增layer

Implement C++ Class

The C++ class of the layer implements the initialization, forward, and backward part of the layer. It needs to derive the base class paddle::Layer, and it needs to override the following functions:

• constructor and destructor.
• init function. It is used to initialize the parameters and settings.
• forward. It implements the forward part of the layer.
• backward. It implements the backward part of the layer.
• prefetch. It is utilized to determine the rows corresponding parameter matrix to prefetch from parameter server. You do not need to override this function if your layer does not need remote sparse update. (most layers do not need to support remote sparse update)

头文件：

namespace paddle {
/**
 * A layer has full connections to all neurons in the previous layer.
 * It computes an inner product with a set of learned weights, and
 * (optionally) adds biases.
 *
 * The config file api is fc_layer.
 */

class FullyConnectedLayer : public Layer {
protected:
  WeightList weights_;
  std::unique_ptr<Weight> biases_;

public:
  explicit FullyConnectedLayer(const LayerConfig& config)
      : Layer(config) {}
  ~FullyConnectedLayer() {}

  bool init(const LayerMap& layerMap, const ParameterMap& parameterMap);

  Weight& getWeight(int idx) { return *weights_[idx]; }

  void prefetch();
  void forward(PassType passType);
  void backward(const UpdateCallback& callback = nullptr);
};
}  // namespace paddle

It defines the parameters as class variables. We use Weight class as abstraction of parameters. It supports multi-thread update. The details of this class will be described in details in the implementations.

• weights_ is a list of weights for the transformation matrices. The current implementation can have more than one inputs. Thus, it has a list of weights. One weight corresponds to an input.
• biases_ is a weight for the bias vector.

The fully connected layer does not have layer configuration hyper-parameters. If there are some layer hyper-parameters, a common practice is to store it in LayerConfig& config, and put it into a class variable in the constructor.

The following code snippet implements the init function.

• First, every init function must call the init function of the base class Layer::init(layerMap, parameterMap);. This statement will initialize the required variables and connections for each layer.
• The it initializes all the weights matrices $W$ . The current implementation can have more than one inputs. Thus, it has a list of weights.(当前layer的输入可能来自多个layer，每个layer对应一个weight)
• Finally, it initializes the bias.

bool FullyConnectedLayer::init(const LayerMap& layerMap,
                               const ParameterMap& parameterMap) {
  /* Initialize the basic parent class */
  Layer::init(layerMap, parameterMap);

  /* initialize the weightList */
  CHECK(inputLayers_.size() == parameters_.size());
  for (size_t i = ; i < inputLayers_.size(); i++) {
    // Option the parameters
    // 输入层的神经元数目
    size_t height = inputLayers_[i]->getSize();
    // 当前层的神经元数目
    size_t width = getSize();

    // create a new weight
    if (parameters_[i]->isSparse()) {
      CHECK_LE(parameters_[i]->getSize(), width * height);
    } else {
      CHECK_EQ(parameters_[i]->getSize(), width * height);
    }
    Weight* w = new Weight(height, width, parameters_[i]);

    // append the new weight to the list
    weights_.emplace_back(w);
  }

  /* initialize biases_ */
  if (biasParameter_.get() != NULL) {
    biases_ = std::unique_ptr<Weight>(new Weight(, getSize(), biasParameter_));
  }

  return true;
}

The implementation of the forward part has the following steps.

• Every layer must call Layer::forward(passType); at the beginning of its forward function.
• Then it allocates memory for the output using reserveOutput(batchSize, size);. This step is necessary because we support the batches to have different batch sizes. reserveOutput will change the size of the output accordingly. For the sake of efficiency, we will allocate new memory if we want to expand the matrix, but we will reuse the existing memory block if we want to shrink the matrix.
• Then it computes $\sum_i W_i x + b$ using Matrix operations。 getInput(i).value retrieve the matrix of the i-th input. Each input is a $batchSize×dim$ matrix, where each row represents an single input in a batch. For a complete lists of supported matrix operations, please refer to paddle/math/Matrix.h and paddle/math/BaseMatrix.h.

• Finally it applies the activation function using forwardActivation();. It will automatically applies the corresponding activation function specifies in the network configuration.

void FullyConnectedLayer::forward(PassType passType) {
  Layer::forward(passType);

  /* malloc memory for the output_ if necessary */
  // batchSize是样本数，size是神经元数目
  int batchSize = getInput().getBatchSize();
  int size = getSize();

  {
    // Settup the size of the output.
    reserveOutput(batchSize, size);
  }

  MatrixPtr outV = getOutputValue();

  // Apply the the transformation matrix to each input.
  for (size_t i = ; i != inputLayers_.size(); ++i) {
    auto input = getInput(i);
    CHECK(input.value) << "The input of 'fc' layer must be matrix";
    i ==  ? outV->mul(input.value, weights_[i]->getW(), , )
           : outV->mul(input.value, weights_[i]->getW(), , );
  }

  /* add the bias-vector */
  if (biases_.get() != NULL) {
    outV->addBias(*(biases_->getW()), );
  }

  /* activation */ {
    forwardActivation();
  }
}

The implementation of the backward part has the following steps.

• backwardActivation() computes the gradients of the activation. The gradients will be multiplies in place to the gradients of the output, which can be retrieved using getOutputGrad().
• Compute the gradients of bias. Notice that we an use biases_->getWGrad() to get the gradient matrix of the corresponding parameter. After the gradient of one parameter is updated, it must call getParameterPtr()->incUpdate(callback);. This is utilize for parameter update over multiple threads or multiple machines.

• Then it computes the gradients of the transformation matrices and inputs, and it calls incUpdate for the corresponding parameter. This gives the framework the chance to know whether it has gathered all the gradient to one parameter so that it can do some overlapping work (e.g., network communication)

void FullyConnectedLayer::backward(const UpdateCallback& callback) {
  /* Do derivation for activations.*/ {
    // 计算本层网络的激活关于本层网络参数的偏导
    backwardActivation();
  }

  if (biases_ && biases_->getWGrad()) {
    // 计算loss函数关于本层网络偏差的梯度
    biases_->getWGrad()->collectBias(*getOutputGrad(), );

    biases_->getParameterPtr()->incUpdate(callback);
  }

  bool syncFlag = hl_get_sync_flag();

  for (size_t i = ; i != inputLayers_.size(); ++i) {
    /* Calculate the W-gradient for the current layer */
    if (weights_[i]->getWGrad()) {
      MatrixPtr input_T = getInputValue(i)->getTranspose();
      MatrixPtr oGrad = getOutputGrad();
      {
        weights_[i]->getWGrad()->mul(input_T, oGrad, , );
      }
    }

    /* Calculate the input layers error */
    MatrixPtr preGrad = getInputGrad(i);
    if (NULL != preGrad) {
      MatrixPtr weights_T = weights_[i]->getW()->getTranspose();
      preGrad->mul(getOutputGrad(), weights_T, , );
    }

    {
      weights_[i]->getParameterPtr()->incUpdate(callback);
    }
  }
}

The prefetch function specifies the rows that need to be fetched from parameter server during training. It is only useful for remote sparse training. In remote sparse training, the full parameter matrix is stored distributedly at the parameter server. When the layer uses a batch for training, only a subset of locations of the input is non-zero in this batch. Thus, this layer only needs the rows of the transformation matrix corresponding to the locations of these non-zero entries. The prefetch function specifies the ids of these rows.

Most of the layers do not need remote sparse training function. You do not need to override this function in this case.

void FullyConnectedLayer::prefetch() {
  for (size_t i = ; i != inputLayers_.size(); ++i) {
    auto* sparseParam =
        dynamic_cast<SparsePrefetchRowCpuMatrix*>(weights_[i]->getW().get());
    if (sparseParam) {
      MatrixPtr input = getInputValue(i);
      sparseParam->addRows(input);
    }
  }
}

Finally, you can use REGISTER_LAYER(fc, FullyConnectedLayer); to register the layer. fc is the identifier of the layer, and FullyConnectedLayer is the class name of the layer.

namespace paddle {
REGISTER_LAYER(fc, FullyConnectedLayer);
}

If the cpp file is put into paddle/gserver/layers, it will be automatically added to the compilation list.

Implement Python Wrapper

Implementing Python wrapper allows us to use the added layer in configuration files. All the Python wrappers are in file python/paddle/trainer/config_parser.py. An example of the Python wrapper for fully connected layer is listed below. It has the following steps:

• Use @config_layer('fc') at the decorator for all the Python wrapper class. fc is the identifier of the layer.

• Implements __init__ constructor function.

  • It first call super(FCLayer, self).__init__(name, 'fc', size, inputs=inputs, **xargs) base constructor function. FCLayer is the Python wrapper class name, and fc is the layer identifier name. They must be correct in order for the wrapper to work.
  • Then it computes the size and format (whether sparse) of each transformation matrix as well as the size.

@config_layer('fc')
class FCLayer(LayerBase):
    def __init__(
            self,
            name,
            size,
            inputs,
            bias=True,
            **xargs):
        super(FCLayer, self).__init__(name, 'fc', size, inputs=inputs, **xargs)
        for input_index in xrange(len(self.inputs)):
            input_layer = self.get_input_layer(input_index)
            psize = self.config.size * input_layer.size
            dims = [input_layer.size, self.config.size]
            format = self.inputs[input_index].format
            sparse = format == "csr" or format == "csc"
            if sparse:
                psize = self.inputs[input_index].nnz
            self.create_input_parameter(input_index, psize, dims, sparse, format)
        self.create_bias_parameter(bias, self.config.size)

In network configuration, the layer can be specifies using the following code snippets. The arguments of this class are:

• name is the name identifier of the layer instance.
• type is the type of the layer, specified using layer identifier.
• size is the output size of the layer.
• bias specifies whether this layer instance has bias.
• inputs specifies a list of layer instance names as inputs.

Layer(
    name = "fc1",
    type = "fc",
    size = ,
    bias = True,
    inputs = [Input("pool3")]
)

You are also recommended to implement a helper for the Python wrapper, which makes it easier to write models. You can refer to python/paddle/trainer_config_helpers/layers.py for examples.

http://doc.paddlepaddle.org/doc/howto/dev/new_layer_en.html

paddle源码解析：

http://wiki.babel.baidu.com/twiki/bin/view/Main/Paddle%E6%BA%90%E7%A0%81%E5%89%96%E6%9E%90--Layer#2.2 backward函数

http://wiki.baidu.com/pages/viewpage.action?pageId=353372756