Neural Network Playground

Visualize and understand how neural networks learn to solve classification problems

Settings

Network Architecture

Problem TypeChoose the type of problem for the neural network to solve

Hidden LayersNumber of layers between input and output

Neurons per LayerNumber of neurons in each hidden layer

Training Parameters

Learning RateSpeed of learning (smaller = more stable but slower)

Activation FunctionFunction used to determine neuron activation

Controls

Epoch

Error

0.000

Accuracy

0.0%

Learning to separate two interleaving spiral patterns. Each spiral represents a different class in this challenging non-linear classification problem.

Neural Network Playground

Settings

Network Architecture

Training Parameters

Controls

Neural Network Structure

Classification Results