A neural network is layers of simple mathematical functions stacked together. Each layer learns increasingly abstract features. It's modelled loosely on the brain, but don't take that analogy too far.

A neural network is a function approximator. Feed it input, it produces output, and you adjust its parameters (weights) to minimise error.

**Structure:**

- **Input layer**: raw features (pixel values, word embeddings)

- **Hidden layers**: learned transformations (the 'magic' happens here)

- **Output layer**: prediction (class probabilities, regression value)

**How it learns:**

Forward pass → calculate loss → backward pass (backpropagation) → update weights.

**'Deep' just means many layers.** More layers = model can learn more complex, hierarchical features.

Neural Networks: Layers of Learnable Logic