Figure 8 | Scientific Reports

Figure 8

From: E-I balance emerges naturally from continuous Hebbian learning in autonomous neural networks

Figure 8

Time evolution of the average effective excitatory (violet) and inhibitory (green) synaptic weights \({\overline{w}}^{(exc)}\) and \({\overline{w}}^{(inh)}\), as defined by (6). The network contains 200 and 200 excitatory and inhibitory neurons. Also shown is the average balanced weight (red), given by \({\overline{w}}^{(exc)}+{\overline{w}}^{(inh)}\). Left: Using Oja’s rule (8). Right: Using the flux rule (2), as for Fig. 6, but this time with the limiting factor G(x) = x0 + x(1 − 2y) replaced by a constant, G → 10. Both approaches fail to produce a balanced synaptic weight distribution.

Back to article page