Reproducibility of biophysical in silico neuron states and spikes from event-based partial histories
Fig 7
Spike train replicability of HH and WB models with event-based input encodings as a function of number of inputs.
van Rossum distance between observed spike trains from the neuron models with and without the event-based input encoding for each of the five HH model parameterizations and the WB model. Shown are the distances for encodings which include the output spike (solid lines) and encodings which do not include output spike (hashed line). When very limited historical stimuli information is available, including the time of the last output spike reduced van Rossum error, but with larger amounts of stimuli history, for all but the baseline case the van Rossum error was lower on models generated without incorporating the last output spike time.