Neural networks power everything from voice assistants to medical diagnosis — but most textbooks either skip the math entirely or bury students in calculus before explaining the basic idea. If you have a class, a project, or an exam coming up and you need to understand how artificial neurons actually compute, this guide cuts straight to the point.

**Neural Networks Explained** is a focused 10–20 page primer that walks you through the full picture: how a single neuron takes inputs, applies weights and a bias, and fires an output; how neurons stack into layers to approximate complex functions; and how a network measures its own mistakes with a loss function and fixes them using gradient descent. The guide then unpacks backpropagation — the chain-rule algorithm that tells every weight in every layer exactly how much it contributed to an error — and closes with a practical survey of CNNs, RNNs, and Transformers so you know which architecture fits which kind of data.

This book is written for high school students in CS or AI electives, college freshmen encountering machine learning for the first time, and anyone who searched for *backpropagation explained simply* and got a Wikipedia page they couldn't parse. Every concept comes with worked numbers, plain-English definitions, and callouts for the misconceptions students most often bring into exams.

Short by design. Read it in one sitting, then walk into class ready.

• Explain what an artificial neuron is and how it computes an output from weighted inputs and an activation function
• Describe how neurons are stacked into layers to form feedforward networks and why depth matters
• Define a loss function and explain how gradient descent uses it to update weights
• Walk through backpropagation at a conceptual level, including the role of the chain rule
• Recognize common architectures (CNNs, RNNs, transformers) and the kinds of problems each one fits

1. 1. What a Neural Network Actually Is
   Introduces neural networks as function approximators built from simple units, and distinguishes them from brains and from traditional programming.
2. 2. Inside a Single Neuron: Weights, Bias, and Activation
   Breaks down the arithmetic of one artificial neuron with a worked numerical example and explains why nonlinear activation functions are essential.
3. 3. Stacking Neurons into Layers
   Shows how neurons combine into input, hidden, and output layers to form feedforward networks, and why depth lets networks represent complex patterns.
4. 4. Learning from Data: Loss and Gradient Descent
   Explains how networks measure their own mistakes with a loss function and adjust weights using gradient descent.
5. 5. Backpropagation: How the Network Knows What to Fix
   Walks through the chain-rule logic that lets a network distribute blame for an error back through every weight in every layer.
6. 6. Beyond the Basics: CNNs, RNNs, and Transformers
   Surveys the main architectures built on top of the basic neuron and matches each to the kind of data it handles best.