1. STDP Learning Rule
Spike-Timing Dependent Plasticity (STDP) is the biological hack that lets neurons strengthen or weaken their connections based on who “spiked” first. Dominate the timing, dominate the weight.
Why it matters: Precise timing means your simulated neural network learns patterns faster, no feeble emotions allowed haha, only raw timing power.
STDP Window Function:
Δw = A₊ · exp(−Δt/τ₊) for Δt > 0 (potentiation)
Δw = −A₋ · exp(Δt/τ₋) for Δt < 0 (depression)
Δw = A₊ · exp(−Δt/τ₊) for Δt > 0 (potentiation)
Δw = −A₋ · exp(Δt/τ₋) for Δt < 0 (depression)
| Parameter | What It Controls | Default Value |
|---|---|---|
| A₊ | Strength boost when post follows pre | 0.1 |
| A₋ | Strength drop when pre follows post | 0.12 |
| τ₊ | Time window for potentiation (ms) | 20 |
| τ₋ | Time window for depression (ms) | 20 |
2. Memristor STDP Emulation
Memristors are tiny resistors with memory, they change conductance based on voltage history. Use them to mimic STDP naturally, without writing a single rule by hand.
Why it matters: Hardware-level plasticity means real-time learning devices that don’t rely on brute-force code loops.
Memristor Conductance Model:
G(t) = G₀ + ΔG · f(V(t), t)
dG/dt = η · V(t) · g(G, V)
G(t) = G₀ + ΔG · f(V(t), t)
dG/dt = η · V(t) · g(G, V)
| Parameter | Role | Default |
|---|---|---|
| G₀ | Starting conductance (µS) | 1 |
| η | Learning rate factor | 0.01 |
| V_th | Voltage threshold to trigger change (V) | 0.5 |
| τ_m | Memristor’s intrinsic time constant (ms) | 10 |