Neural Network

A neural network is a mathematical model loosely inspired by how biological brains work. It consists of layers of "neurons" — simple units that multiply their inputs by adjustable numbers called weights, sum the results, and pass them through an activation function.

How it learns

• The network starts with random weights and makes random predictions.
• For each training example, it measures how wrong its prediction was (the loss).
• Using backpropagation and gradient descent, it nudges every weight slightly in the direction that reduces the error.
• Repeated over thousands of examples and many epochs, the network discovers patterns in the data.

Common uses

• Classification (spam detection, image recognition, fraud detection)
• Prediction and forecasting
• Language understanding (the foundation of modern LLMs like GPT and Claude)

Try it yourself Our Neural Network Playground lets you train a real neural network in your browser — no code required — and watch the weights change as it learns.