Design, wire, and simulate arbitrary neural networks. Explore logic gates, memory circuits, and finite automata.
Build networks from scratch or load predefined templates.
Place input, hidden, and output neurons on the grid, wire them with excitatory (+1) or inhibitory (−1) connections,
and step through the simulation to watch signals propagate with the one-step synaptic delay.
Predefined templates include logic gates, an SR latch, flip-flops, and finite automata that illustrate the
Kleene–McCulloch–Pitts theorem: cyclic M-P networks recognize exactly the regular languages.
Toolbox
Place Neuron
Input
Hidden
Output
Click a type, then click on the canvas to place. Press Esc to cancel. To wire: click a neuron's right port, then click the target neuron.
Load Template
Simulation
Time stept = 0
Inspector
Click a neuron or connection to inspect it.
Network State
No neurons yet.
Load a predefined circuit from the template menu to see its mathematical description here.