Question

我正在尝试修改此代码（请参阅下面的GitHub链接），以便我可以使用自己的数据并使用相同的功能集预测多个标签。

https://github.com/tensorflow/tensorflow/blob/master/tensorflow/examples/tutorials/input_fn/boston.py

当我一次使用一个标签时，我可以使用它。但是当我尝试创建一个包含多个标签的张量时，我遇到了问题。有什么建议吗？

我修改的LABELS和input_fn如下所示：

LABELS = ["Label1", "Label2", "Label3"]

def input_fn(data_set):
  feature_cols = {k: tf.constant(len(data_set), shape=[data_set[k].size, 1]) for k in FEATURES}
  labels_data = []
  for i in range(0, len(data_set)):
    temp = []
    for label in LABELS:
        temp.append(data_set[label].values[i])
    labels_data.append(temp)
  labels = tf.constant(labels_data, shape=[len(data_set), len(LABELS)])
  return feature_cols, labels

这是我收到的错误消息的结尾：

  File "/usr/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/dnn.py", line 175, in _dnn_model_fn
return head.head_ops(features, labels, mode, _train_op_fn, logits)
  File "/usr/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py", line 403, in head_ops
head_name=self.head_name)
  File "/usr/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py", line 1358, in _training_loss
loss_fn(logits, labels),
  File "/usr/lib/python2.7/site-packages/tensorflow/contrib/learn/python/learn/estimators/head.py", line 330, in _mean_squared_loss
logits.get_shape().assert_is_compatible_with(labels.get_shape())
  File "/usr/lib/python2.7/site-packages/tensorflow/python/framework/tensor_shape.py", line 735, in assert_is_compatible_with
raise ValueError("Shapes %s and %s are incompatible" % (self, other))
ValueError: Shapes (118, 1) and (118, 3) are incompatible

Answer 1

来自this link好像你必须指定3个类（标签）的数量：

regressor = tf.contrib.learn.DNNRegressor(feature_columns=feature_cols,
                                        n_classes=3,
                                        hidden_units=[10, 10],
                                        model_dir="/tmp/boston_model")

n_classes 值应该设置为2而不是3，看看哪个有用

Answer 2

从here开始，您必须更改此参数：

label_dimension: Dimension of the label for multilabels. Defaults to 1.

所以这应该有效：

regressor = tf.contrib.learn.DNNRegressor(feature_columns=feature_cols,
                                        label_dimension=3,
                                        hidden_units=[10, 10],
                                        model_dir="/tmp/boston_model")

具有张量流的多个标签

2 个答案: