Neural Network STDP Simulation
About this simulation:
This simulation models spike-timing dependent plasticity (STDP) in a neural network with excitatory and inhibitory neurons using mean-field theory. The key equations being simulated are:
Key Equations
1. Firing Rate Dynamics
τr
d𝐫⃗
dt
= -𝐫 + 𝕎𝐫 + 𝕎X𝐫X
Where 𝐫 = [rE, rI] are the excitatory and inhibitory firing rates
2. Connectivity Matrices
𝕎 = (
(NE-1)wEE -NIwEI
NEwIE -(NI-1)wII
)
𝕎X = Nx (
wEX 0
0 wIX
)
𝐫X = [aE, aI] are the external drive parameters
3. STDP Weight Evolution
d𝑤EE
dt
= (τSTDP/τwEE) rE(rE - b)
d𝑤EI
dt
= (τSTDP/τwEI) rI(rE - b)
Synaptic weights evolve based on pre- and post-synaptic activity with threshold b
Typical parameter combinations to try:
- σ = 0.1, aE = 20, wEE = 10, wEI = 10, Time = 10s (standard)
- σ = 0.3, aE = 30, wEE = 10, wEI = 10, Time = 20s (higher activity)
- σ = 0.0, aE = 15, wEE = 10, wEI = 10, Time = 50s (no noise, longer)
- σ = 0.5, aE = 25, wEE = 10, wEI = 10, Time = 30s (high noise)