Question

我正在玩R中的玩具神经网络，试图学习keras语法。我对此很陌生，所以觉得我可能会错过一些简单的东西。

这是我的X矩阵的前几行：

train_x <- structure(c(619, 502, 699, 850, 645, 822, 0, 0, 0, 0, 0, 0, 0, 
0, 0, 1, 1, 0, 0, 0, 0, 0, 1, 1, 42, 42, 39, 43, 44, 50, 2, 8, 
1, 2, 8, 7, 0, 159660.8, 0, 125510.82, 113755.78, 0, 1, 3, 2, 
1, 2, 2, 1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 0, 1, 101348.88, 113931.57, 
93826.63, 79084.1, 149756.71, 10062.8), .Dim = c(6L, 11L), .Dimnames = 
list(
c("1", "3", "4", "5", "6", "7"), c("CreditScore", "GeographyGermany", 
"GeographySpain", "GenderMale", "Age", "Tenure", "Balance", 
"NumOfProducts", "HasCrCard", "IsActiveMember", "EstimatedSalary"
)))

这是y的前几个：

train_y <- c(1L, 1L, 0L, 0L, 1L, 0L)

这是我正在努力训练的模型

class <- keras_model_sequential()

class %>%
layer_dense(units = 6, input_shape = 11, 
          activation = 'relu', 
          kernel_initializer = "uniform") %>%
  layer_dropout(0.2) %>%
  layer_dense(units = 6, 
          activation = 'relu', 
          kernel_initializer = 'uniform') %>%
  layer_dropout(0.2) %>%
  layer_dense(units = 1, activation = 'sigmoid',
          kernel_initializer = 'uniform')

class %>%
  compile(optimizer = 'adam', 
      loss = 'binary_crossentropy', 
      metrics = 'accuracy')
class %>%
  fit(train_x, train_y, batch_size = 10, 
  epochs = 10)

y_hat <- predict_classes(class, test_x)
table(y_hat)

我已经尝试过搞乱一堆这些东西，但似乎无法让模型预测除了所有零或全零之外的任何东西。我想我错过了一些简单的......有什么建议吗？

为什么这不适用于Keras？

0 个答案: