Visual AI · Experiment 01

Training

Training can be better understood by seeing how it acts on the data.

Explore training

During this training, the two hidden neurons in the neural network draw their lines (or mathematical functions) over the data, trying to separate the red data from the blue data. Training ends when the neurons find the best way to separate the data. The final state of the neural network becomes the model.

Watch the network learn

The points stay fixed. The decision boundaries are what change.

Before training: the boundaries do not yet match the pattern in the examples.