Winner-Take-All Network
Lateral inhibition: competition selects one winner; click neurons to boost
Network
N neurons
6
Inhib strength
2.00
Self-excit
1.20
Noise σ
0.10
Each neuron inhibits all others. Winner = neuron with highest net input. Inhib strength controls competition sharpness.
Inputs
Base input
1.50
Signal-to-noise
0.50
τ (time const)
5.0
Threshold
0.50
Click neural circles above to boost that neuron's input. SNR controls how distinct the winning signal is.
Decision
—
WTA implements soft argmax when competition weak, hard argmax when strong. Models attention, action selection, categorical perception.