Interactive machine learning lab

Neural Network Demo

Train a tiny classifier, explore how it learns, and test predictions on your own inputs.

2 inputs · 2 classes · 4 examples

Experiment setup

Change the sample point or training settings, then run the model.

Prediction input

Input 1

Input 2

Training settings

Learning rate 0.1

Epochs 1000

Batch size 64

Optimizer

Training runs right here in your browser.

Training not started yet.

Live training loss

Ready when you are — press Train model to start.

Training progress0%

Decision boundary

Background color shows the model's predicted class across the input space. Dots are training examples; the outlined dot is your current input.

Class 1 Class 0

3D probability surface

The surface height and color represent the network's output probability as both inputs change. Drag to rotate, or focus the chart and use the arrow keys.

Lower output Higher output