Stochastic
- 21.5 seconds of the input was on stochastic input.
- Run on the earlier parameters, the model gets 92 of the 94
spikes in the clamped period, and 151 of 200 on the stochastic.
- A quick analysis shows there are missing spikes under
high input (theta too high) and mising spikes under low input
(Decay too low).
- A change to D = 1.12 (from 1.1) and Threshold at 2.2 (from 2.6).
- This gave 14 spikes at 300 (real 11), but 183 of 200 on stocahstic.
- I aligned the spikes (in XL) with the biological data and
of the 288, 26 had two model spikes aligned with a biological one.
- I took the first of these, and the average diffence was 16.3 ms.
- So, the model gets 260/292 (89 percent) spikes right within an average of 1.5 cycles.
- Comparing on a blind second run, it gets 261/288 (91 percent) spikes
right within an average of 15.3 ms.