Building a simple multilayer perceptron model in PyTorch to do handwritten digit classification. This is done very similary to the logistic regression example, but instead of going straight to a 10-d vector representing our output predictions, we might first transform to a 500-d vector with a "hidden" layer, then to the output of dimension 10.
One more important thing to consider is that We typically include nonlinearities between layers of a neural network. There's a number of reasons to do so. For one, without anything nonlinear between them, successive linear transforms (fully connected layers) collapse into a single linear transform, which means the model isn't any more expressive than a single layer. On the other hand, intermediate nonlinearities prevent this collapse, allowing neural networks to approximate more complex functions.
The MNIST dataset is very popular machine learning dataset, consisting of 70000 grayscale images of handwritten digits, of dimensions 28x28.
We split the dataset into two separate groups:
Number of MNIST training examples: 60000
Number of MNIST test examples: 10000
So, the goal is to predict which digit is in each image in the testing set. Then, we compute an accuracy by seeing how many we got correct!