The Neural Network Lab

A neural network learns to make predictions by training on data. Here you see the whole process from the inside.

1
Explore the calculations. Follow how each neuron multiplies, sums and activates.
2
Train the network. Choose a scenario and let the network learn from a dataset.
3
Rebuild the network. Add neurons and hidden layers and examine how the loss is affected.

You begin in Calculation mode, where every neuron works with simple whole numbers. Then switch to a scenario and let the network train itself.

AI-aktiviteter

The Neural Network Lab

Follow the calculations step by step, train on data, and make predictions.

Svenska
Scenario
Architecture
Input variables
Activation
Press Train the model to begin.
Concepts
Input Neuron Weight Bias (+b) Activation Activation function ReLU Sigmoid Tanh Hidden layer Output Loss Accuracy Forward pass Epoch Prediction

Training data

The first columns show real-world values. The same values then follow, normalized to the range 0 to 1, which is what the network sees.