I built an image recognizing spreadsheet to better understand Convolutional Neural Networks. The sheet can recognize the letters X, T, and L. Creating it really helped me understand the details, especially how a deep convolution layer can build off the previous layers.
I created a two-part series that walks through building that spreadsheet.
Part 1 - https://www.youtube.com/watch?v=8QfFtUcuYCU
Part 2 - https://www.youtube.com/watch?v=pH0s0e62GT8
The fully implemented sheet is at https://bit.ly/CNNSheet3 if you just want to jump to the end.
If you are just getting into neural networks, maybe look at my previous series on building and training a traditional neural network in a spreadsheet:
Top comments (0)