machine-learning - 参数无效错误预期 begin[0] = 0

Question

我目前正在开发一个神经网络，我获得了所有数据和代码，将图像馈送到 CNN 进行训练。但是，在训练过程中，对于第一张图像，会弹出错误，代码如下。

    def convolutional_neural_network(x):
    weights = {'W_conv1':tf.Variable(tf.random_normal([5,5,1,32])),
               'W_conv2':tf.Variable(tf.random_normal([5,5,32,64])),
               'W_fc':tf.Variable(tf.random_normal([7*7*64,1024])),
               'out':tf.Variable(tf.random_normal([1024, n_classes]))}

    biases = {'b_conv1':tf.Variable(tf.random_normal([32])),
               'b_conv2':tf.Variable(tf.random_normal([64])),
               'b_fc':tf.Variable(tf.random_normal([1024])),
               'out':tf.Variable(tf.random_normal([n_classes]))}

    x = tf.reshape(x, shape=[-1, 28, 28, 1])

    conv1 = tf.nn.relu(conv2d(x, weights['W_conv1']) + biases['b_conv1'])
    conv1 = maxpool2d(conv1)

    conv2 = tf.nn.relu(conv2d(conv1, weights['W_conv2']) + biases['b_conv2'])
    conv2 = maxpool2d(conv2)

    fc = tf.reshape(conv2,[-1, 7*7*64])
    fc = tf.nn.relu(tf.matmul(fc, weights['W_fc'])+biases['b_fc'])
    fc = tf.nn.dropout(fc, keep_rate)

    output = tf.matmul(fc, weights['out'])+biases['out']
    print("hi")
    return output


def shuffle_unison(images, labels):
    shuffleLabel = []
    shuffleImage = []
    shuffleVector = []
    for i in range(0, len(images)-1):
        shuffleVector.append(i)
    random.shuffle(shuffleLabel)
    for i in range(0, len(shuffleVector)-1):
        shuffleImage.append(images[shuffleVector[i]])
        shuffleLabel.append(labels[shuffleVector[i]])
    return shuffleImage, shuffleLabel





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

    hm_epochs = 10
    # step 4: Batching

    with tf.Session() as sess:
        init = tf.initialize_all_variables()
        sess.run(init)
        tf.train.start_queue_runners()
        #array of strings and corresponding values
        image_list, label_list = readImageLables()
        for epoch in range(hm_epochs):
            epoch_loss = 0
            #shuffle every epoch
            shuffle_image_list, shuffle_label_list = shuffle_unison(image_list, label_list)
            sampleList = ['/home/sciencefair/Desktop/OrchardData/MachineLearningTesting/RottenOranges/result1.jpg']
            for i in range(0,7683):
                #filename_queue = tf.train.string_input_producer(sampleList)
                file_contents =  tf.read_file(shuffle_image_list[i])
                image = tf.image.decode_jpeg(file_contents, channels=1)
                resized_image = tf.image.resize_images(image, [28,28])
                #image_batch, label_batch = tf.train.batch([resized_image, shuffle_label_list[i]], batch_size=batch_size) # does train.batch take individual images or final tensors
                #if(i>batch_size):
                    #print(label_batch.eval())
                a = tf.reshape(resized_image,[1, 784])
                print(a.eval())
                _, c = sess.run([optimizer, cost], feed_dict={x: tf.reshape(resized_image,[1, 784]).eval(), y: shuffle_label_list[i]})
                epoch_loss += c
                print("ok")

            print('Epoch', epoch, 'completed out of',hm_epochs,'loss:',epoch_loss)
        sess.close()

堆栈跟踪看起来像这样

    Caused by op 'Slice_1', defined at:
  File "revisednet.py", line 128, in <module>
    train_neural_network(x)
  File "revisednet.py", line 87, in train_neural_network
    cost = tf.reduce_mean( tf.nn.softmax_cross_entropy_with_logits(prediction,y) )
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/nn_ops.py", line 670, in softmax_cross_entropy_with_logits
    labels = _flatten_outer_dims(labels)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/nn_ops.py", line 472, in _flatten_outer_dims
    array_ops.shape(logits), [math_ops.sub(rank, 1)], [1])
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/array_ops.py", line 431, in slice
    return gen_array_ops._slice(input_, begin, size, name=name)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/gen_array_ops.py", line 2234, in _slice
    name=name)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/op_def_library.py", line 749, in apply_op
    op_def=op_def)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/ops.py", line 2380, in create_op
    original_op=self._default_original_op, op_def=op_def)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/ops.py", line 1298, in __init__
    self._traceback = _extract_stack()

InvalidArgumentError (see above for traceback): Expected begin[0] == 0 (got -1) and size[0] == 0 (got 1) when input.dim_size(0) == 0
     [[Node: Slice_1 = Slice[Index=DT_INT32, T=DT_INT32, _device="/job:localhost/replica:0/task:0/cpu:0"](Shape_2, Slice_1/begin, Slice_1/size)]]

此错误似乎源于数据与 softmax 函数发生冲突。但是我完全不知道是什么导致了这个问题。

Answer 1

我遵循了本教程:Sentdex,First pass through Data w/ 3D ConvNet构建 3D CNN 并得到与您相同的错误。

出现此错误是因为我的输入数据的标签向量的维度(例如，Sentdex 的训练数据中第一个标签向量的位置是 train_data[0][1])应该是与 n_classes 相同的数字，在教程中为 2。

在我错误的尝试中，我只是使用二进制值0或1来表示它，其维度是1，而应该是2。所以tf.nn.softmax_cross_entropy_with_logits()函数被混淆了标签向量的大小错误。

尝试将标签向量的维度扩展为等于您的n_classes。