public List <Tensor> GetLossList(ProcessData data)
{
Tensor[] lossList = new Tensor[this.Count];
List <Tensor> pushes = this.GetAllOutputs(data.Data);
Tensor loss;
//这里对softmax层进行了特别对待处理，应该如何统一化？
if (layers[layerCount - 1].Sign == LayerSign.SoftMaxLayer)
{
loss = data.Label - pushes[this.layerCount - 1];
}
else
{
loss = LossFunc.Loss(data.Label, pushes[this.layerCount - 1]);;
}
Tensor nextWeight = TensorBuilder.FromMatrix(Matrix <double> .Build.DenseIdentity(data.Label[0, 0].ColumnCount));
LayerSign nextType = LayerSign.Nothing;
for (int i = this.layerCount - 1; i >= 2; i--)
{
loss = layers[i].ComputeLoss(loss, pushes[i - 1], nextWeight, nextType);
lossList[i] = loss;
nextWeight = layers[i].Weight;
nextType = layers[i].Sign;
}
loss = layers[1].ComputeLoss(loss, pushes[0], nextWeight, nextType);
lossList[1] = loss;
nextWeight = layers[1].Weight;
nextType = layers[1].Sign;
loss = layers[0].ComputeLoss(loss, data.Data, nextWeight, nextType);
lossList[0] = loss;
return(lossList.ToList());
}
/// <summary>
/// 获取梯度列表，同时为了简化运算同时输出损失列表
/// </summary>
/// <param name="data"></param>
/// <returns>第一个为梯度列表，第二个为损失列表</returns>
public Tuple <Tensor[], Tensor[]> GetGradientList(ProcessData data)
{
Tensor[] gradientList = new Tensor[this.Count];
Tensor[] lossList = new Tensor[this.Count];
List <Tensor> pushes = this.GetAllOutputs(data.Data);
Tensor loss;
//这里对softmax层进行了特别对待处理，应该如何统一化？
if (layers[layerCount - 1].Sign == LayerSign.SoftMaxLayer)//反射是否开销太大了？
{
loss = data.Label - pushes[this.layerCount - 1];
}
else
{
loss = LossFunc.Loss(data.Label, pushes[this.layerCount - 1]);
}
Tensor nextWeight = TensorBuilder.FromMatrix(Matrix <double> .Build.DenseIdentity(data.Label[0, 0].RowCount));
LayerSign nextType = LayerSign.Nothing;
for (int i = this.layerCount - 1; i >= 2; i--)
{
(gradientList[i], loss) = layers[i].GetGradient(loss, pushes[i - 1], nextWeight, nextType);
lossList[i] = loss;
//loss = layers[i].ComputeLoss(loss, pushes[i - 1], nextWeight);
nextWeight = layers[i].Weight;
nextType = layers[i].Sign;
}
(gradientList[1], loss) = layers[1].GetGradient(loss, pushes[0], nextWeight, nextType);
lossList[1] = loss;
//loss = layers[1].ComputeLoss(loss, pushes[0], nextWeight);
nextWeight = layers[1].Weight;
nextType = layers[1].Sign;
(gradientList[0], lossList[0]) = layers[0].GetGradient(loss, data.Data, nextWeight, nextType);
return(new Tuple <Tensor[], Tensor[]>(gradientList, lossList));
}
/// <summary> /// 计算梯度（同时返回损失，减少后续计算开销） /// </summary> /// <param name="nextLoss"></param> /// <param name="preOutput"></param> /// <param name="nextWeight"></param> /// <returns>第一个为梯度，第二个为损失</returns> public abstract Tuple <Tensor, Tensor> GetGradient(Tensor nextLoss, Tensor preOutput, Tensor nextWeight, LayerSign nextType);
public abstract Tensor ComputeLoss(Tensor nextLoss, Tensor preOutput, Tensor nextWeight, LayerSign nextType);
public Layer(NoiseType _type, LayerSign _sign)
{
type = _type;
sign = _sign;
}
