Fig. 2: The signal reconstruction task.
From: A multisynaptic spiking neuron for simultaneously encoding spatiotemporal dynamics
a The overview of the spiking autoencoder for simple signal reconstruction. b Examples of original and reconstructed temporal signals (TP = 300) based on LIF (T = 300) and MSF (T = 300) neurons. The upper part shows sinusoidal temporal signals with varying frequencies and amplitudes, while the lower part shows square wave temporal signals with varying frequencies, where intensity is either 0 or 1. c Intensity distributions of original and reconstructed spatial signals (T = 1) based on LIF (T = 1), rate-coded LIF (T = 4), and MSF (T = 1) neurons. d Relative reduction percentage in image and video reconstruction errors achieved by MSF neurons compared to baseline LIF neurons. e Visualization comparison of reconstructed images and videos.
