具有自定义约束的Keras Load_model不起作用

Question

当尝试使用自定义约束类加载保存的模型时，出现问题。

我的班级如下：

class WeightsOrthogonalityConstraint (Constraint):
def __init__(self, encoding_dim, weightage = 1.0, axis = 0):
    self.encoding_dim = encoding_dim
    self.weightage = weightage
    self.axis = axis

def weights_orthogonality(self, w):
    if(self.axis==1):
        w = K.transpose(w)
    if(self.encoding_dim > 1):
        m = K.dot(K.transpose(w), w) - K.eye(self.encoding_dim)
        return self.weightage * K.sqrt(K.sum(K.square(m)))
    else:
        m = K.sum(w ** 2) - 1.
        return m

def __call__(self, w):
    return self.weights_orthogonality(w)

def get_config(self):
    return {
        'encoding_dim': self.encoding_dim,
        'weightage': self.weightage,
        'axis': self.axis
    }

并且，我像这样使用此类：

encoder = Dense(encoding_dim, activation="linear", 
                #input_shape=(input_dim,), 
                use_bias = True, 
                kernel_regularizer=WeightsOrthogonalityConstraint(encoding_dim, weightage=1., axis=0), 
                kernel_constraint=UnitNorm(axis=0))

然后，我这样称呼load_model：

autoencoder=load_model('anomaly-detection_Fully.h5',custom_objects={'WeightsOrthogonalityConstraint': WeightsOrthogonalityConstraint(16)})

但是，发生错误

    File "C:\ProgramData\Anaconda3\lib\site-packages\keras\utils\generic_utils.py", line 154, in deserialize_keras_object
    return cls(**config['config'])
TypeError: __call__() missing 1 required positional argument: 'w'

我怎么会通过“ w”？