Home▸AI & Machine Learning▸Neural Network Signal Sweep (2D)

Neural Network Signal Sweep (2D)

A 2D companion to the 3D Interactive Neural Network Simulation: the same 5-layer perceptron (6-10-12-10-4) sweeps a forward-propagation wave through its sparsified connections layer by layer, but here you also get a flat wiring diagram and a live per-layer firing-rate strip chart plus a spike raster you cannot see in the 3D scene.

AI & Machine Learning2DEasy60 FPS⇄ 3D version

This is the 2D counterpart to the 3D Interactive Neural Network Simulation. It keeps the exact same fixed 5-layer perceptron (6-10-12-10-4 neurons, ~55% of inter-layer connections wired), the same round-robin forward-propagation wave, and the same activation-speed / wave-frequency / brightness controls — but lays the network out flat as a wiring diagram and adds a live per-layer firing-rate strip chart and pulse counters the 3D scene has no room to show. Neither version trains on data; both are pure visualisations of how a signal sweeps through a fixed feedforward architecture.

⚙ Under the hood

A 2D companion to the 3D Interactive Neural Network Simulation: the same fixed 6-10-12-10-4 perceptron with about 55% of inter-layer connections wired sweeps a forward-propagation wave through its layers in round-robin order, but laid out as a flat wiring diagram with a live per-layer firing-rate strip chart and pulse/cycle counters the 3D scene has no room to show.

neural networkforward propagationactivation wavesignal flowperceptroncanvas 2d

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