11RanNeural nets & vision

Clothing images with a dense network

I flattened 28 by 28 Fashion-MNIST images and trained a 300-100-10 dense network with SGD for 30 epochs, holding out 5,000 training images for validation.

Original tutorial by Aman Kharwal: Classification with Neural Networks (opens in a new tab)

What it printed in 2022

Numbers read from the saved notebook outputs. Cell numbers count every cell in the notebook, Markdown included, starting at 1.
Output	Value	Cell #
Validation accuracy after epoch 30	88.9%	9
Training accuracy after epoch 30	91.8%	9
Network parameters	266,610	6

Cell # counts every cell from the top of the notebook, Markdown included, starting at 1.

88.9% is validation accuracy. The 10,000-image test set was only used for five example predictions.

The data

Fashion-MNIST, downloaded through Keras.

What I'd do differently

Score the test set once at the end and report that number.
Plot training and validation curves. Validation loss had stopped improving while training loss kept falling.
Look at the confusion matrix. Shirts, T-shirts and pullovers are where a dense network struggles.