Execute the code while reading Deep Learning created from scratch. I will write down the places where it is jammed.
-Windows10 -Ubuntu on Windows -python3
numpy was already included, but matplotlib was not included, so I installed it with a command.
$ sudo apt-get install python3-matplotlib
show () cannot be used I need to introduce Xming http://qiita.com/makky0620/items/e31edc90f22340d791ff
Download Xming from the following site https://ja.osdn.net/projects/sfnet_xming/
Xming-6-9-0-31-setup.exe Xming-fonts-7-7-0-10-setup.exe 2 I ran the file and installed it.
When you hover the mouse over the icon in the lower right bar, Xming server: 0.0 is displayed (as per the reference site)
Enter in the environment variable.
$ export DISPLAY=localhost:0.0
$ xclock
The clock was displayed successfully.
It is now displayed by show ().
1.6.2 legend () seems to be a function that gives a legend
？ The keyboard was US, probably because I installed Xming.
1.6.3 An error occurred when loading and displaying the image
Traceback (most recent call last):
File "img.py", line 5, in <module>
plt.imshow(img)
File "/usr/lib/python3/dist-packages/matplotlib/pyplot.py", line 2881, in imshow
ax = gca()
File "/usr/lib/python3/dist-packages/matplotlib/pyplot.py", line 803, in gca
ax = gcf().gca(**kwargs)
File "/usr/lib/python3/dist-packages/matplotlib/pyplot.py", line 450, in gcf
return figure()
File "/usr/lib/python3/dist-packages/matplotlib/pyplot.py", line 423, in figure
**kwargs)
File "/usr/lib/python3/dist-packages/matplotlib/backends/backend_tkagg.py", line 79, in new_figure_manager
return new_figure_manager_given_figure(num, figure)
File "/usr/lib/python3/dist-packages/matplotlib/backends/backend_tkagg.py", line 87, in new_figure_manager_given_figure
window = Tk.Tk()
File "/usr/lib/python3.4/tkinter/__init__.py", line 1854, in __init__
self.tk = _tkinter.create(screenName, baseName, className, interactive, wantobjects, useTk, sync, use)
_tkinter.TclError: no display name and no $DISPLAY environment variable
I thought, but I forgot ʻexport DISPLAY = localhost: 0.0`. Every time you restart the terminal, you have to reconfigure the environment. Write in .bashrc.
3.6.1
It says that mnist.py is in the dataset directory, but what is the dataset directory?
It said that you should include the Anaconda distribution, but is it related to not including it?
There was a link to GitHub from the book introduction page on O'Reilly's site.
https://github.com/oreilly-japan/deep-learning-from-scratch
to download.
$ git clone https://github.com/oreilly-japan/deep-learning-from-scratch.git
Images cannot be displayed using PIL.
You can save the file with pil_img.save ("output.png ").
It seems that it can be displayed by inserting something called Imagemagick.
http://d.hatena.ne.jp/kimihito/20120508/1336461904
$ sudo apt-get install imagemagick
An error occurred that pickle was undefined because it was copied. ʻImport pickle`.
4.4.1 Hyperparameters are different from neural network parameters and seem to be adjusted manually. I wonder if this parameter can also be acquired automatically.
4.4.2 I found a (don't care) mistake in the text. It was confirmed that the same was true for the 1st and 6th prints.
P. 111
\>>> net = simpleNet()
\>>> print(net.W) #Weight parameter
[[ 0.47355232, 0.9977393 , 0.84668094],
[ 0.85557411, 0.03563661, 0.69422093]])
\>>>
\>>> x = np.array([0.6, 0.9])
\>>> p = net.predict(x)
\>>> print(p)
[ 1.13282549 0.66052348 1.20919114]```
The above interactive code is included, but the final print result is numerically `[1.05414809 0.63071653 1.1328074]`.
Since net.W is generated by random numbers, I wonder if it was executed once and net.predict (x) was copied to the manuscript of the book, then net.W was regenerated and net.W was copied to the manuscript of the book.
lambda is a syntax for creating anonymous functions.
`f = lambda w: net.loss(x, t)`
When
#### **`python`**
```def f(x)
return net.loss(x, t)```
Is probably equivalent.
Recommended Posts