Module 1 · Introduction to Deep Learning

OBJECTIVE: understand what deep learning is, how a neural network learns, and the tools used to build one — before we start projects. WHAT WE COVER: • What deep learning is and how it differs from machine learning • How a neural network (ANN) is built: neurons, weights, layers, activations • How it learns: loss, gradient descent, backpropagation, optimisers • How to stop overfitting: validation, dropout, early stopping • The main network types and what each is for: ANN, CNN, RNN/LSTM, Transformers • The two main frameworks: PyTorch and TensorFlow/Keras AFTER THIS MODULE YOU CAN: • Explain deep learning in simple words and say when to use it • Read a small network and explain what each layer does • Train your first neural network in Keras and in PyTorch and check its accuracy • Choose the right type of network for images, tables, sequences or text MINI EXERCISE: a first clothing classifier on Fashion-MNIST (no deployment yet — the projects do that).