In addition to implementing and training a neural network from scratch, this notebook will also cover a few other important topics. The reason for this is to try to help you answer some more profound or theoretical questions you may still have about machine learning in general, and about neural networks specifically. These are:
- see equivalency between logistic regression and single perceptron
- understand optimization and why loss functions are used (rather than accuracy)
- utilize functionality of PyTorch (for applying chain rule and for optimization)