Examples
MNIST Digits

MNIST Digit Recognition

Multi-class classification using evolved neural networks

Watch as NEAT evolves a neural network to recognize handwritten digits (0-9) from the classic MNIST dataset. Each image is 28x28 pixels (784 inputs), with 10 output classes.

Performance Warning

This network has 784 input neurons and 10 outputs. It may run slowly in the browser and the network visualization is disabled due to complexity.

Test Set Results

Correct Predictions

of 28000 total

Incorrect Predictions

misclassified digits

Error Rate

0.0%

lower is better

Neural Network

Network Visualization Disabled

784 input + 10 output neurons makes visualization impractical

Error

0.0

Fitness

0.0

Iteration

Network

784-0-10

input - hidden - output

Dataset Information

784

Input Pixels (28x28)

Output Classes (0-9)

28K

Training Samples

Kaggle

Data Source

Performance Warning

Test Set Results

Neural Network

Dataset Information

NEAT Settings