Skip to main content
. Author manuscript; available in PMC: 2017 May 4.
Published in final edited form as: Neuron. 2016 May 4;90(3):471–482. doi: 10.1016/j.neuron.2016.04.014

Figure 6. Spike inference without training data.

Figure 6

a) Schematic illustrating the setup: The algorithms are trained on all cells from three datasets (here: all but the GCaMP dataset) and evaluated on the remaining dataset (here: the GCaMP dataset), testing how well it generalizes to settings it has not seen during training.

b) Correlation (mean± 2 SEM for repeated measure designs) between the true spike rate and the inferred spike density function for a subset of the algorithms (see legend for color code) evaluated on each of the four different datasets collected under anesthesia/ex-vivo (with n=16, 31, 19 and 9, respectively), trained on the remaining three. Markers above bars show the result of a Wilcoxon sign rank test between the STM model and its closest competitor (see Methods, * denotes P<0.05, ** denotes P<0.01). The evaluation was performed in bins of 40 ms.

c) Information gained about the true spike train by observing the calcium trace performing the generalization analysis described in a).