ANNTrainer {Microsoft.VisualBasic.MachineLearning.NeuralNetwork}

.NET clr documentation

ANNTrainer

Description

Declare

            
# namespace Microsoft.VisualBasic.MachineLearning.NeuralNetwork
export class ANNTrainer extends IterationReporter`1 {
   # [0,1]之间,建议设置一个[0.3,0.6]之间的值, 这个参数表示被随机删除的节点的数量百分比,值越高,则剩下的神经元节点越少
   dropOutRate: double;
   ErrorThreshold: double;
   MinError: double;
   # 最终得到的训练结果神经网络
   NeuronNetwork: Network;
   # 是否对训练样本数据集进行选择性的训练，假若目标样本在当前所训练的模型上面
   #  所计算得到的预测结果和其真实结果的误差足够小的话，目标样本将不会再进行训练
   Selective: boolean;
   TrainingSet: TrainingSample[];
   TrainingType: TrainingType;
   # 对Neuron.Gradient的剪裁限制阈值，小于等于零表示不进行剪裁，默认不剪裁
   Truncate: double;
   # 训练所使用到的经验数量,即数据集的大小size
   XP: integer;
}

        

.NET clr type reference tree

1. this class extends from IterationReporter`1 class: IterationReporter`1
2. use by property member NeuronNetwork: Network
3. use by property member TrainingSet: TrainingSample
4. use by property member TrainingType: TrainingType

---