Home▸Networks & Graph Theory▸Neural Network Playground: 2D

🕸️ Neural Network Playground: 2D

2D neural network playground with decision heat map, network diagram and live backpropagation.

Networks & Graph Theory2DModerate60 FPS📱 Mobile-adapted⇄ 2D version

Details

CategoryNetworks & Graph Theory

RenderingCanvas 2D

DifficultyModerate

Target FPS60

ControlsPick a dataset (circle, XOR, spiral, moons, blobs), set samples and noise, choose hidden layers, neurons and activation, tune learning rate, L2 decay, steps per frame and batch size, drag, add or erase data points, click hidden neurons to ablate them, and hover to probe the network.

neural networksbackpropagationmultilayer perceptronactivation functionsclassificationartificial intelligence

Related simulations

Force-Directed GraphFruchterman-Reingold spring layout for Erdős-Rényi, Barabási-Albert and Watts-Strogatz…

Minimum Spanning TreeStep-by-step animation of Kruskal's and Prim's MST algorithms on a weighted graph. Toggle…

Network ResilienceBarabási–Albert scale-free and Erdős–Rényi random networks under targeted hub attack vs… ⊞ All 177 in Networks & Graph Theory →

⚙ Under the hood

Interactive 2D neural network trained live by backpropagation: a decision heat map with draggable data points beside a network diagram whose weights, activations and neurons you can inspect and ablate.

neural networksbackpropagationmultilayer perceptronactivation functionsclassificationartificial intelligence

2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install