Table 2.

Breakdown of classification errors for the GLM random network

Connection type	Errors	FP	FN	ND
Excitatory	7 300	62 %	38 %	15 %
Inhibitory	0
Unconnected	7 300	38 %	62 %	—
Total errors	0.73 %

The column “Errors” shows the absolute number of incorrectly classified connections belonging to each class. The columns “FP” and “FN” show the percentage of false positives and false negatives of this number accordingly. The column “ND” is the percentage of misclassified connections in violation of the Dale’s law (i.e. an inhibitory neuron is assigned an outgoing excitatory connection, or vice versa), which was not enforced for this reconstruction. The last row shows the percentage of erroneously classified connections of the total number of possible connections (N ²=10⁶ for N=1000)