Exploring the Usefulness of Formal Concept Analysis for Robust Detection of Spatio-temporal Spike Patterns in Massively Parallel Spike Trains

Haemmerlé, Ollivier; Stapleton, Gem; Yegenoglu, Alper; Torre, Emiliano; Faron Zucker, Catherine; Grün, Sonja; Quaglio, Pietro; Endres, Dominik

doi:10.1007/978-3-319-40985-6_1

Contribution to a book

FZJ-2016-03545

Exploring the Usefulness of Formal Concept Analysis for Robust Detection of Spatio-temporal Spike Patterns in Massively Parallel Spike Trains

Haemmerlé, O. (Editor) ; Stapleton, G. (Editor) ; Faron Zucker, C. (Editor) ; Yegenoglu, A.FZJ* ; Quaglio, P.FZJ* ; Torre, E.FZJ* ; Grün, S. (Corresponding author)FZJ* ; Endres, D. (Corresponding author)

2016
Springer International Publishing Cham
ISBN: 978-3-319-40984-9, 978-3-319-40985-6 (electronic)

Graph-Based Representation and Reasoning / Haemmerlé, Ollivier (Editor) ; Cham : Springer International Publishing, 2016, Chapter 1 ; ISSN: 0302-9743=1611-3349 ; ISBN: 978-3-319-40984-9=978-3-319-40985-6 ; doi:10.1007/978-3-319-40985-6 Cham : Springer International Publishing, Lecture Notes in Computer Science 9717, 3 - 16 (2016) [10.1007/978-3-319-40985-6_1]

This record in other databases:

Please use a persistent id in citations: doi:10.1007/978-3-319-40985-6_1

Abstract: The understanding of the mechanisms of information processing in the brain would yield practical impact on innovations such as brain-computer interfaces. Spatio-temporal patterns of spikes (or action potentials) produced by groups of neurons have been hypothesized to play an important role in cortical communication [1]. Due to modern advances in recording techniques at millisecond resolution, an empirical test of the spatio-temporal pattern hypothesis is now becoming possible in principle. However, existing methods for such a test are limited to a small number of parallel spike recordings. We propose a new method that is based on Formal Concept Analysis (FCA, [11]) to carry out this intensive search. We show that evaluating conceptual stability [18] is an effective way of separating background noise from interesting patterns, as assessed by precision and recall rates on ground truth data. Because of the scaling behavior of stability evaluation, our approach is only feasible on medium-sized data sets consisting of a few dozens of neurons recorded simultaneously for some seconds. We would therefore like to encourage investigations on how to improve this scaling, to facilitate research in this important area of computational neuroscience.

Classification: