Spiking Neural Network — STDP

Spike-Timing-Dependent Plasticity: Δw = A₊·e^(−Δt/τ₊) if pre→post, −A₋·e^(Δt/τ₋) if post→pre

Mean weight
Output firing rate
Time (ms)
STDP is a Hebbian plasticity rule: synapses strengthen when pre fires before post (causal), weaken otherwise. Slight LTD dominance (A₋ > A₊) stabilizes weight distributions. Neurons use leaky integrate-and-fire (LIF) dynamics. Raster: each row = one neuron (top 20 input, bottom output). Weight heatmap below.