The MNIST dataset contains 60000 pictures of handwritten digits in 28x28 pixels.
- In MNIST.ipynb, the first notebook, I define a few functions to visualize the digits.
- In MNIST_2.ipynb, I do a quick comparaison of different classifiers on the dataset (without doing any hyperparameter tuning).
- In MNIST_3.ipynb, I apply Principal Component Analysis to the dataset, to reduce the dimensionnality of the features. I also use PCA to visualize the data in 2 dimensions. I also use t-SNE for visualization.