However, as the HRI algorithm scans only those sections of the recording indicated by the LRI search, then the entire LRI-HRI detector is compromised by the failing in the LRI.
Using the LRI-HRI detector program, no difference could be found between the implanted trace and its shuffled surrogates (Fig. 6).
This shortcoming is not so much a problem in the HRI detection part of the algorithm as the detector window is matched to the length of the synaptic events (between 20 msec and 100 msec).
Mokeichev et al. (2007) and Ikegaya et al. (2004) used essentially the same detector program, called here the LRI-HRI program, to find repeating patterns of membrane potential fluctuations in cortical intracellular recordings (Fig. 2).
The actual values produced in the final analysis from HRI do not suffer from this defect since the detector window is matched to the width of the individual PSC (20 msec) or PSP (100 msec).
Unfortunately, there can be many false positives from this 1st pass in the detector algorithm such that many candidates never pass the threshold for gaining HRI analysis.
This detector differed from the previous in two ways: (1) putative repeats, to be carried over to HRI analysis, are selected based on the onset times of PSPs; (2) the algorithm that scans the putative repeats, creating T values, does not measure every 100 msec interval, but rather measures each interval as indicated by the onset times of the PSPs (Fig. 7).
Organizer, Algorithms for HRI Workshop, Paris 2016.
Organizer, Planning for HRI Workshop, RSS 2016.
Organizer, Collaborative Manipulation Workshop, HRI 2013.
This work has been supported by the Hydraulics Research Institute (HRI).
