Visual AI · Experiment 02

Improved training

For a neural network to create a more precise decision boundary during training, the network may need more neurons and more hidden layers.

Explore training

During this training, we can observe how a larger neural network architecture, with more neurons, can separate the data in a less linear and more curved way. This allows it to create a tighter decision boundary.

Watch the network learn

Training reshapes the decision boundary to better fit the examples.

Before training: the decision boundary does not yet fit the examples well.