Skip to main content
. Author manuscript; available in PMC: 2009 Nov 25.
Published in final edited form as: Neural Comput. 2008 Jun;20(6):1495–1511. doi: 10.1162/neco.2007.10-06-350

Figure 6.

Figure 6

Spike misassignment and the performances of the metrics. The spike trains have been corrupted by a random relabelling of the spikes, each spike has a λ probability of being swapped over to the other spike train in the response. For both the new and ARMV metrics, h is plotted against λ for the SP and LL metrics, as well as the metric lying half way between these: α = 0.5 for the new metrics and k = 1 for the ARMV metrics. For each value of λ, the value of h is averaged over 20 runs. The runs are all for zero mixing parameter, a = 0, where the LL metrics are at their greatest advantage over the SP metric. The SP distances are invariant under the relabelling of spikes.