Fig. 7.

Performance of BOTM with respect to errors in the template set on benchmark 2 (H). a The performance of BOTM as a function of the number of templates used for classification. A random subset of templates, excluding the target template was deleted from the set of all templates generated by the initial spike sorting before the BOTM was run. Colored bands indicate one standard deviation. b Influence of noise in the templates on BOTM performance. Left panel: For this analysis a random subset of all available spikes for each neuron was used to compute the template. Right panel: A template is shown on all four electrodes, once for using all available spikes (blue line) to compute the template and once for using only two spikes (red line). For this analysis the same data set as in Fig. 4 was used. Sensitivity is not shown since it was not affected by the errors on the template set