Deep Learning & Neural Networks

What you'll learn

• Explain neurons, activations, layers and why depth lets networks learn non-linear patterns.
• Describe how loss functions, backpropagation and optimisers train a network, and check gradients.
• Write a complete PyTorch training loop with tensors, modules, data loaders and evaluation mode.
• Diagnose and fix training problems such as vanishing gradients, overfitting and bad learning rates.
• Explain and build convolutional networks, and explain embeddings, attention and transformer blocks.
• Apply transfer learning, debug models systematically, and save and serve them correctly.

Syllabus

Neural Network Foundations

1. What Deep Learning Is
2. Neurons and Activation Functions
3. Why Hidden Layers Matter: The XOR Problem

How Networks Learn

1. Loss Functions
2. Gradient Descent and Backpropagation
3. Automatic Differentiation in PyTorch
4. Optimisers: SGD, Momentum and Adam

The PyTorch Workflow

1. Tensors and Shapes
2. Modules and the Training Loop
3. Mini-Batches, Epochs and Devices

Making Training Work

1. Vanishing Gradients, Normalisation and Residual Connections
2. Regularisation: Dropout, Weight Decay and Augmentation
3. Learning-Rate Schedules and Early Stopping

Convolutional Neural Networks

1. The Convolution Operation
2. Building a Small CNN
3. Why Convolutions Are Efficient

Sequences, Embeddings and Attention

1. Embeddings: Tokens as Vectors
2. Recurrent Networks and Their Limits
3. Self-Attention

Transformers, Transfer Learning and Generative Models

1. The Transformer Block
2. Transfer Learning and Fine-Tuning
3. Generative Models at a Glance

Debugging, Deploying and a Checklist

1. Debugging Neural Networks
2. Saving, Loading and Serving Models
3. A Deep Learning Project Checklist