Keras 开关功能错误

Question

我正在使用 tensorflow 作为后端为 keras 创建自定义损失函数。为此，我想尝试类似于此处显示的内容：以 y_true 将取决于 y_pred 的方式自定义 Keras 的损失函数

但是，在运行脚本时（取自 Van）：

import theano
from keras import backend as K
from keras.layers import Dense
from keras.models import Sequential

def customized_loss(y_true, y_pred):
    loss = K.switch(K.equal(y_true, -1), 0, K.square(y_true-y_pred))
    return K.sum(loss)

if __name__ == '__main__':
    model = Sequential([ Dense(3, input_shape=(4,)) ])
    model.compile(loss=customized_loss, optimizer='sgd')

    import numpy as np
    x = np.random.random((1, 4))
    y = np.array([[1,-1,0]])

    output = model.predict(x)
    print output
    # [[ 0.47242549 -0.45106074  0.13912249]]
    print model.evaluate(x, y)  # keras's loss
    # 0.297689884901
    print (output[0, 0]-1)**2 + 0 +(output[0, 2]-0)**2 # double-check
    # 0.297689929093

我收到以下错误： AttributeError: 'int' object has no attribute 'get_shape'

然后我尝试使用 tf.where 代替（如 Van 的评论中所见），但是我收到了这个错误：ValueError: Shapes must be equal rank, but are 0 and 2 for 'loss/dense_1_loss/Select' (op: '选择') 输入形状：[?,?], [], [?,3]。

任何帮助，将不胜感激

编辑：完整的堆栈跟踪是：对于第一种情况（K.switch）：

Using TensorFlow backend.
2018-02-21 14:47:22.907033: I C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\36\tensorflow\core\platform\cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
Traceback (most recent call last):
  File "temp.py", line 19, in <module>
    model.compile(loss=customized_loss, optimizer='sgd')
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\models.py", line 806, in compile
    **kwargs)
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\engine\training.py", line 860, in compile
    sample_weight, mask)
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\engine\training.py", line 460, in weighted
    score_array = fn(y_true, y_pred)
  File "temp.py", line 10, in customized_loss
    loss = K.switch(K.equal(y_true, -1), 0, K.square(y_true-y_pred))
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\backend\tensorflow_backend.py", line 2674, in switch
    expr_ndim = ndim(then_expression)
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\backend\tensorflow_backend.py", line 590, in ndim
    dims = x.get_shape()._dims
AttributeError: 'int' object has no attribute 'get_shape'

第二个错误（使用 tf.where 而不是 K.switch）：

Using TensorFlow backend.
2018-02-21 14:49:31.651045: I C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\36\tensorflow\core\platform\cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
Traceback (most recent call last):
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 686, in _call_cpp_shape_fn_impl
    input_tensors_as_shapes, status)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\errors_impl.py", line 473, in __exit__
    c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.InvalidArgumentError: Shapes must be equal rank, but are 0 and 2 for 'loss/dense_1_loss/Select' (op: 'Select') with input shapes: [?,?], [], [?,3].

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "temp.py", line 19, in <module>
    model.compile(loss=customized_loss, optimizer='sgd')
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\models.py", line 806, in compile
    **kwargs)
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\engine\training.py", line 860, in compile
    sample_weight, mask)
  File "C:\Users\shai\Anaconda3\lib\site-packages\keras\engine\training.py", line 460, in weighted
    score_array = fn(y_true, y_pred)
  File "temp.py", line 12, in customized_loss
    loss = tf.where(K.equal(y_true, -1), 0.0, K.square(y_true - y_pred))
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\ops\array_ops.py", line 2441, in where
    return gen_math_ops._select(condition=condition, t=x, e=y, name=name)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\ops\gen_math_ops.py", line 3987, in _select
    "Select", condition=condition, t=t, e=e, name=name)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 787, in _apply_op_helper
    op_def=op_def)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 2958, in create_op
    set_shapes_for_outputs(ret)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 2209, in set_shapes_for_outputs
    shapes = shape_func(op)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 2159, in call_with_requiring
    return call_cpp_shape_fn(op, require_shape_fn=True)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 627, in call_cpp_shape_fn
    require_shape_fn)
  File "C:\Users\shai\Anaconda3\lib\site-packages\tensorflow\python\framework\common_shapes.py", line 691, in _call_cpp_shape_fn_impl
    raise ValueError(err.message)
ValueError: Shapes must be equal rank, but are 0 and 2 for 'loss/dense_1_loss/Select' (op: 'Select') with input shapes: [?,?], [], [?,3].

Answer 1

而不是使用普通整数作为参数来switch传递兼容的张量，例如使用以下方法创建它zeros_like：

loss = K.switch(K.equal(y_true, -1), K.zeros_like(y_true), K.square(y_true-y_pred))