Theano Faltung: TypeError: conv2d() hat mehrere Werte für das Argument 'input'

Question

Ich versuche, eine "doppelte" Ebene mit Faltung erst dann max Pooling zu erstellen. Das Netzwerk soll 20x20 Eingangsbilder erhalten und eine Klasse von [0,25] ausgeben. Beim Versuch, die Funktionen zu erstellen, erhalte ich den Fehler TypeError: conv2d() got multiple values for argument 'input' beim Aktivieren der Faltungspooling-Schicht.

class ConvPoolLayer: 
    conv_func = T.nnet.conv2d 
    pool_func = max_pool_2d 

    def __init__(self, image_shape, n_feature_maps, act_func, 
       local_receptive_field_size=(5,5), pool_size=(2,2), 
       init_weight_func=init_rand_weights, init_bias_weight_func=init_rand_weights): 
     """ 
     Generate a convolutional and a subsequent pooling layer with one bias node for each channel in the pooling layer. 
     :param image_shape: tuple(batch size, input channels, input rows, input columns) where 
      input_channels = number of feature maps in upstream layer 
      input rows, input columns = output size of upstream layer 
     :param n_feature_maps: number of feature maps/filters in this layer 
     :param local_receptive_field_size: = size of local receptive field 
     :param pool_size: 
     :param act_func: 
     :param init_weight_func: 
     :param init_bias_weight_func: 
     """ 
     self.image_shape = image_shape 
     self.filter_shape = (n_feature_maps, image_shape[1]) + local_receptive_field_size 
     self.act_func = act_func 
     self.pool_size = pool_size 
     self.weights = init_weight_func(self.filter_shape) 
     self.bias_weights = init_bias_weight_func((n_feature_maps,)) 
     self.params = [self.weights, self.bias_weights] 
     self.output_values = None 

    def activate(self, input_values): 
     """ 
     :param input_values: the output from the upstream layer (which is input to this layer) 
     :return: 
     """ 
     input_values = input_values.reshape(self.image_shape) 
     conv = self.conv_func(
      input=input_values, 
      image_shape=self.image_shape, 
      filters=self.weights, 
      filter_shape=self.filter_shape 
     ) 
     pooled = self.pool_func(
      input=conv, 
      ds=self.pool_size, 
      ignore_border=True 
     ) 
     self.output_values = self.act_func(pooled + self.bias_weights.dimshuffle('x', 0, 'x', 'x')) 

    def output(self): 
     assert self.output_values is not None, 'Asking for output before activating layer' 
     return self.output_values 


def test_conv_layer(): 
    batch_size = 10 
    input_shape = (20, 20) 
    output_shape = (26,) 
    image_shape = (batch_size, 1) + input_shape # e.g image_shape = (10, 1, 20, 20) 
    n_feature_maps = 10 
    convpool_layer = ConvPoolLayer(image_shape, n_feature_maps, T.nnet.relu) 

    x = T.fmatrix('X') 
    y = T.fmatrix('Y') 

    convpool_layer.activate(x) 


test_conv_layer() 

Answer 1

Das Problem ist, dass Sie conv_func() als eine Methode für Ihre Klasse ConvPoolLayer() festlegen. Also, wenn Sie tun:

conv = self.conv_func(input=input_values, 
         image_shape=self.image_shape, 
         filters=self.weights, 
         filter_shape=self.filter_shape) 

Python, hinter der Szene dies tut:

conv = ConvPoolLayer.conv_func(self, input=input_values, 
           image_shape=self.image_shape, 
           filters=self.weights, 
           filter_shape=self.filter_shape) 

und seit input der erste Parameter ist dann mehrere Werte für sie liefern.

Sie können dies vermeiden, indem Sie die Methoden in static Einwickeln() wie folgt aus:

conv_func = staticmethod(T.nnet.conv2d) 

oder aus __init__ die conv_func Attribut festlegen. Beachten Sie, dass Sie das gleiche Problem für pool_func haben werden.