video thumbnail 9:08
Neural Networks (E03: data and evaluation – python)

2018-06-14

[public] 85.3K views, 3.34K likes, 37.0 dislikes audio only

channel thumbSebastian Lague

In this episode we load our training data and evaluate how accurate the network is with random weights.

Download the data here (see readme for more info on loading):

https://github.com/SebLague/Mnist-data-numpy-format

Code:

https://github.com/SebLague/Neural-Network-python

I owe a lot to this excellent online book on neural networks:

http://neuralnetworksanddeeplearning.com/

Support the creation of more tutorials:

https://www.patreon.com/SebastianLague


get the arrays from the data file
/youtube/video/d9hLNUzLBYI?t=99.849998
print out training images dot shape
/youtube/video/d9hLNUzLBYI?t=146.75
interpret the values as grayscale
/youtube/video/d9hLNUzLBYI?t=237.82899
set the input layer size to 784
/youtube/video/d9hLNUzLBYI?t=269.14001
pass the training images into the prediction method
/youtube/video/d9hLNUzLBYI?t=281.01999
print out the accuracy of the network
/youtube/video/d9hLNUzLBYI?t=367.98999
compare these predictions to the labels
/youtube/video/d9hLNUzLBYI?t=390.19
Sebastian Lague Hello, on this channel I explore how to create stuff out of code. I hope you find some of it interesting!
/youtube/channel/UCmtyQOKKmrMVaKuRXz02jbQ
Support these videos on Patreon patreon.com
https://www.patreon.com/SebastianLague
Introduction to Game Development (Unity and C#) by Sebastian Lague
/youtube/video/_cCGBMmMOFw
Blender Character Creation by Sebastian Lague
/youtube/video/DiIoWrOlIRw