Beginner-friendly · no scary math
Learn how neural networks think — by playing with them
Inputs are sensors. Weights are control knobs. Activation is the firing threshold. Drag, slide and tune your way to understanding.
Lessons
What Is a Neural Network?
Meet the tiny decision machine. Inputs are sensors, weights are control knobs, and activation is the firing threshold that makes a neuron say 'yes!'
Weights, Learning Rate & Gain
How a network actually improves: measure the error, then take a step. Meet the step size (learning rate), the stretch knob (gain), and the scoreboard (loss).
Activation Functions: ReLU, Sigmoid & Tanh
Three favourite firing rules, with the real formulas and a hand-worked example for each.
How Networks Learn: Five Training Paradigms
Training is a search for good weights. Tour the five big families — supervised, unsupervised, reinforcement, neuroevolution and self-supervised — with real numbers and one shared target.
Backpropagation & Gradient Descent in Depth
The actual training machinery: forward passes, loss functions, the slope of the loss, the update rule with worked iterations, blame assignment by the chain rule, and vanishing gradients.
Train an Agent with the Playground
Lessons 4 and 5 covered how learning works. This one is about setting a goal that learning can actually reach: fitness, shaped rewards, and how to read a training curve. Your lab is the Playground.
Building Brains That Remember
Our agent only sees the world through instant snapshots, and that is why it gets stuck. Meet memory and recurrence: a short note passed forward, a scratchpad hidden state, and why some maps stay impossible until the agent can remember.