基于theano的卷积神经网络实现中遇到的问题

No module named theano.tensor.nnet

在使用Anaconda Navigator-jupyter notebook运行下面的python代码时，出现上述问题。

import theano
from theano.tensor.nnet import conv
import theano.tensor as T

解决方法：

在Anaconda – Environments中安装需要的theano组件。

注意：有时会因为网络问题，导致部分安装失败，可以多次尝试。

UserWarning: downsample module has been moved

在运行下面代码的时候，出现如下错误提示：

from theano.tensor.signal import downsample

解决方法：将上述代码改为：

from theano.tensor.signal import pool
# from theano.tensor.signal import downsample

同时，对于使用了下面代码的语句，也需要做一些改动。

pool_out = downsample.max_pool_2d(input, maxpool_shape, ignore_border=True)
# 改为：
pool_out = pool.pool_2d(input, maxpool_shape, ignore_border=True)

ImportError: No module named logistic_sgd

在运行下面的代码时，提示上面的错误信息：

from logistic_sgd import LogisticRegression, load_data
from mlp import HiddenLayer

解决方法：

在下面的网页中，复制相应的代码，并在jupyter运行文件的相同目录下，创建logistic_sgd.py和mlp.py文件：

• http://deeplearning.net/tutorial/code/logistic_sgd.py
• http://deeplearning.net/tutorial/code/mlp.py

文件目录如下：

python NameError: global name ‘__file__’ is not defined

在运行logistic_sgd.ipynb文件时，出现上述错误提示。出现错误的代码如下：

if data_dir == "" and not os.path.isfile(dataset):
       # Check if dataset is in the data directory.
       new_path = os.path.join(
           os.path.split('__file__')[0], # __file__
           "..",
           "data",
           dataset
       )
       if os.path.isfile(new_path) or data_file == 'mnist.pkl.gz':

解决方法：

I solved it by treating file as a string, i.e. put "__file__" (together with the quotes!) instead of __file__. This works fine for me:

wk_dir = os.path.dirname(os.path.realpath('__file__'))

Python: IOError: [Errno 2] No such file or directory

在python加载图片文件的时候，出现如上错误，代码如下：

img1 = Image.open(open('//home//rachel//Documents//ZJU_Projects//DL//Dataset//rachel.jpg'

解决方法：根据自己需要加载的文件路径，改为如下代码：

# open random image of dimensions 639x516
img_file = open('C:\\Users\\yitian.z\\Documents\\CNNDataset\\littledog.jpg',  'rb')
img = Image.open(img_file) #C:\Users\yitian.z\Documents\CNNDataset

ValueError: total size of new array must be unchanged

在运行下面代码时，出现错误：

img1 = Image.open(open('//home//rachel//Documents//ZJU_Projects//DL//Dataset//rachel.jpg')) 
width1,height1 = img1.size 
img1 = numpy.asarray(img1, dtype = 'float32')/256. # (height, width, 3) 

# put image in 4D tensor of shape (1,3,height,width) 
img1_rgb = img1.swapaxes(0,2).swapaxes(1,2).reshape(1,3,height1,width1) #(3,height,width)

解决方法：

# open random image of dimensions 639x516
img_file = open('C:\\Users\\yitian.z\\Documents\\CNNDataset\\littledog.jpg',  'rb')
img = Image.open(img_file) #C:\Users\yitian.z\Documents\CNNDataset
# img.show()
 
# print img.size
width, height = img.size
print width, height
 
# dimensions are (height, width, channel)
img = numpy.asarray(img, dtype = 'float64')/256. # (height, width, 3)
 
# put image in 4D tensor of shape (1, 3, height, width)
# img1_rgb = img1.swapaxes(0, 2)
# img1_rgb = img1_rgb.swapaxes(1, 2)
# img1_rgb = img1_rgb.reshape(1, 3, height1, width1) # (3, height, width)
cc = img.transpose(2, 0, 1)
img_ = img.transpose(2, 0, 1).reshape(1, 3, height, width)
filtered_img = f(img_)