Fig. 11.

Misclassification error rate as a function of the number of sampled neurons. The reconstructions on the partial datasets were performed with the parameters and data presented in Section 3.1 (black dots). The full network consists of N=1000 neurons, partial datasets consist of the spike trains of a smaller number of randomly selected neurons (maintaining the ratio of excitatory and inhibitory cells). Annotations give the percentage of neurons sampled to the full network size. The chance level MER (20) was calculated for p=0.2, f _e=0.8 and f _i=0.2 (MER₀=33.28 %; dashed line). The synaptic weights were classified using k-means as described in Section 3.3