BP Artificial Neural Network with three layer. Using this network to recognize MNIST handwritten digits.
- python2.7
- numpy
pip install numpy
- pillow
pip install pillow
- GUI need PyQt4
https://www.riverbankcomputing.com/software/pyqt/download
you can install from binary package here PyQt4-4.11.4-gpl-Py2.7-Qt4.8.7-x32.exe
- Mnist Database
http://yann.lecun.com/exdb/mnist/
- train-labels-idx1-ubyte
- train-images-idx3-ubyte
- t10k-labels-idx1-ubyte
- t10k-images-idx3-ubyte
- copy
data.json
from best dictionary to root dictionary(withfile_tools.py
same dictionary). - copy
t10k-images.idx3-ubyte
file to img dictionary and runimg.py
(ensure PIL installed) to get MNIST images. - run
gui.pyw
(ensure PyQt4 installed),you can drag MNIST image to GUI program. - enjoy and happy coding.