Coincidence Detection Using Spiking Neurons with Application to Face Recognition

Table 1

Tabular description of experimental setup as suggested in [36].

	Supervised STDP	Prediction
Model summary
Neuron model	Zero Order Spike Response Model (SRM0) with OSTP
Synaptic model	-shaped PSP
Synaptic function
Input connectivity	Feed forward, with each input neuron connected to each image pixels
Parameters
Weight,	Trained, start at constant 4	5
Synaptic time constant,	3 ms	Trained, start at constant 1 ms
Learning time constant, and	ms	none
Threshold,	9 mV	5.1 mV
No-spike reset potential,	0 mV	2 mV
Spike reset potential,	0 mV	−2 mV
Learning constant,	0.3	1.0
Learning constant,	0.1	none
OSTP discrete interval	10	10
Positive sample per subject,	All available	All available
Negative sample per subject,	1 (random selection)	None
Input model
Type	Direct conversion
Details	Input spike time is equivalent to normalized value of image pixels; Each neuron population is attached to each local patch