Skip to main content

Neural Network Fundamentals

The Intuition (no jargon)​

Imagine you're trying to decide whether a photo contains a cat. After seeing thousands of examples, your brain starts picking up patterns like: pointy ears, whiskers, face shape → likely a cat.

A neural network does something similar. It learns patterns from data instead of being explicitly programmed with rules.

A “network” means many simple computing units (neurons) arranged in layers, passing information forward. Each neuron slightly transforms the signal before passing it on.

Going Deeper: The Actual Mechanics​

Structure​

A neural network typically has three types of layers:

  • Input layer — receives raw data (pixels, numbers, tokens, etc.)
  • Hidden layers — where feature learning happens
  • Output layer — produces final prediction (e.g., cat / not cat)

Each connection has a weight, which controls how strongly one neuron influences another.

👉 Training = learning the best values for these weights.

The Forward Pass​

Data flows from input → output.

At each neuron:

  1. Multiply inputs by weights
  2. Add them up (plus bias)
  3. Apply an activation function

Mathematically: