% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@INPROCEEDINGS{Stella:885610,
author = {Stella, Alessandra and Bouss, Peter and Palm, Günther and
Grün, Sonja},
title = {{R}eoccurring spatio-temporal spike patterns in parallel
spike trains},
school = {RWTH Aachen},
reportid = {FZJ-2020-03962},
year = {2020},
abstract = {We designed the statistical method SPADE [1,2,3] to detect
ms-precise significant spatio-temporal spike patterns (STPs)
across neurons in electrophysiological data. SPADE first
discretizes the spike trains by binning and clipping into
0-1 sequences. These parallel sequences are then mined for
candidate patterns through Frequent Itemset Mining [4]. The
patterns found and their counts are evaluated for
significance using surrogates.A classical choice for
surrogate generation is uniform dithering (UD) [5], i.e.,
uniform independent displacement of each spike around its
original position. In this work, however, we evidence that
UD surrogate spike trains do not conserve the ISI
distribution and the regularity of the original data.
Moreover, we observe that the combination of spike train
binarization and UD causes a reduction of the surrogates’
spike count, yielding an overestimation of the significance
of mined patterns (leading to false positives). Thus, our
objective is to investigate alternatives to UD for the
significance test of spatio-temporal patterns in spike train
data.Therefore, we present five (novel as well as
established) surrogate techniques and contrast them to UD:
uniform dithering with refractoriness, ISI-dithering,
joint-ISI dithering [6,8], spike train shifting [7,8] and
bin shuffling. We compare these surrogate techniques in
terms of the different statistical features of the original
spike trains that they preserve, and we test them on
randomly generated data reflecting the statistics of our
experimental data.We see that only the UD leads to a large
number of false positive patterns. Moreover, when analyzing
experimental data from monkey pre-/motor cortex we detect
many behavior-related patterns and find that all methods
besides UD lead to very similar results in terms of the
found patterns and their compositions, underlining the
robustness of the extracted patterns.AcknowledgementsThe
project is funded by the Helmholtz Association Initiative
and Networking Fund (ZT-I-0003), by Human Brain Project HBP
Grant No. 785907 (SGA2 and SGA3), and by RTG2416
MultiSenses-MultiScales (DFG).References Torre et al.
(2016), 10.1523/jneurosci.4375-15.2016 Quaglio et al.
(2017), 10.3389/fncom.2017.00041 Stella et al. (2019),
10.1016/j.biosystems.2019.104022 Picado-Muiño et al.
(2013), 10.3389/fninf.2013.00009 Date et al. (1998) Gerstein
(2004) Pipa et al. (2008), 10.1007/s10827-007-0065-3 Louis
et al. (2010), $10.1007/978-1-4419-5675-0_17$ Brochier et
al. (2018), 10.1038/sdata.2018.55},
month = {Sep},
date = {2020-09-29},
organization = {Bernstein Conference 2020, Online
(Online), 29 Sep 2020 - 2 Oct 2020},
subtyp = {After Call},
cin = {INM-6 / IAS-6 / INM-10},
cid = {I:(DE-Juel1)INM-6-20090406 / I:(DE-Juel1)IAS-6-20130828 /
I:(DE-Juel1)INM-10-20170113},
pnm = {571 - Connectivity and Activity (POF3-571) / HAF -
Helmholtz Analytics Framework (ZT-I-0003) / HBP SGA2 - Human
Brain Project Specific Grant Agreement 2 (785907) / HBP SGA3
- Human Brain Project Specific Grant Agreement 3 (945539) /
GRK 2416 - GRK 2416: MultiSenses-MultiScales: Neue Ansätze
zur Aufklärung neuronaler multisensorischer Integration
(368482240) / PhD no Grant - Doktorand ohne besondere
Förderung (PHD-NO-GRANT-20170405)},
pid = {G:(DE-HGF)POF3-571 / G:(DE-HGF)ZT-I-0003 /
G:(EU-Grant)785907 / G:(EU-Grant)945539 /
G:(GEPRIS)368482240 / G:(DE-Juel1)PHD-NO-GRANT-20170405},
typ = {PUB:(DE-HGF)24},
doi = {10.12751/NNCN.BC2020.0231},
url = {https://juser.fz-juelich.de/record/885610},
}