Tensorflow中的反捲積

實現函數：tf.nn.conv2d_transpose(input_tensor, filter_weights_tensor, output_shape, strides, padding) input_tensor: 代表的是輸入的tensor，默認的size是[batch_size, w, h, channel] filter_weight_tensor: 代表卷積運算時filter參數的t

>>阅读原文<<