Tensorflow错误：无效参数：形状必须是向量。

Question

我正在使用Kaggle上的泰坦尼克号数据进行tensorflow测试：https://www.kaggle.com/c/titanic 这是我试图从Sendex实现的代码：https://www.youtube.com/watch?v=PwAGxqrXSCs&index=46&list=PLQVvvaa0QuDfKTOs3Keq_kaG2P55YRn5v#t=398.046664

import tensorflow as tf
import cleanData
import numpy as np

train, test = cleanData.read_and_clean()
train = train[['Pclass', 'Sex', 'Age', 'Fare', 'Child', 'Fam_size', 'Title', 'Mother', 'Survived']]

# one hot
train['Died'] = int('0')
train["Died"][train["Survived"] == 0] = 1

print(train.head())

n_nodes_hl1 = 500
n_classes = 2
batch_size = 100

# tf graph input
x = tf.placeholder("float", [None, 8])
y = tf.placeholder("float")

def neural_network_model(data):

    hidden_layer_1 = {'weights':tf.Variable(tf.random_normal([8, n_nodes_hl1])),
                      'biases':tf.Variable(tf.random_normal(n_nodes_hl1))}

    output_layer = {'weights':tf.Variable(tf.random_normal([n_nodes_hl1, n_classes])),
                    'biases':tf.Variable(tf.random_normal([n_classes]))}

    l1 = tf.add(tf.matmul(data, hidden_layer_1['weights']), hidden_layer_1['biases'])
    l1 = tf.nn.relu(l1)

    output = tf.matmul(l1, output_layer['weights']) + output_layer['biases']

    return output

def train_neural_network(x):
    prediction = neural_network_model(x)
    cost = tf.reduce_mean( tf.nn.softmax_cross_entropy_with_logits(prediction,y))
    optimizer = tf.train.AdamOptimizer().minimize(cost)

    desired_epochs = 10

    with tf.Session() as sess:
        sess.run(tf.initialize_all_variables())

        for epoch in range(desired_epochs):
            epoch_loss = 0
            for _ in range(int(train.shape[0])/batch_size):
                x_epoch, y_epoch = train.next_batch(batch_size)
                _, c = sess.run([optimizer, cost], feed_dict= {x:x, y:y})
                epoch_loss += c
            print('Epoch', epoch, 'completed out of', desired_epochs, 'loss:', epoch_loss)

        correct = tf.equal(tf.argmax(prediction,1), tf.argmax(y,1))
        accuracy = tf.reduce_mean(tf.cast(correct, 'float'))
        print('Training accuracy:', accuracy.eval({x:x, y:y}))



train_neural_network(x)

当我运行代码时，出现了一个错误，显示为："W tensorflow/core/framework/op_kernel.cc:909] Invalid argument: shape must be a vector of {int32,int64}, got shape []"。

有没有解决方法？我在Github上看到了tensorflow的代码，显然该库不能将pandas dataframe作为输入。

- Clement

---

请给我们提供一个最简程序示例，清晰地显示出您在哪一行得到了错误信息。 - John Zwinck

Answer 1

我认为错误在这一行上： ````

我认为错误在这一行上：

````

    hidden_layer_1 = {'weights': tf.Variable(tf.random_normal([8, n_nodes_hl1])),
                      'biases': tf.Variable(tf.random_normal(n_nodes_hl1))}

tf.random_normal()函数的shape参数必须是一个整数的一维向量（或列表、数组）。对于变量'biases'，你传递了一个单独的整数n_nodes_hl1。解决方法很简单，只需将该参数放入列表中即可：

    hidden_layer_1 = {...,
                      'biases': tf.Variable(tf.random_normal([n_nodes_hl1]))}

- mrry

Answer 2

得到了不同代码相同的错误，这个修复了：

x = tf.reshape(x, [])

- EliadL