When brain and behavior disagree: Tackling systematic label noise in EEG data with machine learning

Anne K. Porbadnigk, Nico Görnitz, Claudia Sannelli, Alexander Binder, Mikio Braun, Marius Kloft, Klaus Muller

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Conventionally, neuroscientific data is analyzed based on the behavioral response of the participant. This approach assumes that behavioral errors of participants are in line with the neural processing. However, this may not be the case, in particular in experiments with time pressure or studies investigating the threshold of perception. In these cases, the error distribution deviates from uniformity due to the heteroscedastic nature of the underlying experimental set-up. This problem of systematic and structured (non-uniform) label noise is ignored when analysis are based on behavioral data, as is being done typically. Thus, we run the risk to arrive at wrong conclusions in our analysis. This paper proposes a remedy to handle this crucial problem: we present a novel approach for a) measuring label noise and b) removing structured label noise. We show its usefulness for an EEG data set recorded during a standard d2 test for visual attention.

Original languageEnglish
Title of host publication2014 International Winter Workshop on Brain-Computer Interface, BCI 2014
PublisherIEEE Computer Society
DOIs
Publication statusPublished - 2014 Jan 1
Event2014 International Winter Workshop on Brain-Computer Interface, BCI 2014 - Gangwon, Korea, Republic of
Duration: 2014 Feb 172014 Feb 19

Other

Other2014 International Winter Workshop on Brain-Computer Interface, BCI 2014
CountryKorea, Republic of
CityGangwon
Period14/2/1714/2/19

Fingerprint

Electroencephalography
Learning systems
Labels
Brain
brain
learning
remedies
experiment
Processing
Experiments

Keywords

  • Applied Cognitive Neuroscience
  • EEG
  • Label Noise
  • Machine Learning
  • Unsupervised Learning

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Human Factors and Ergonomics

Cite this

Porbadnigk, A. K., Görnitz, N., Sannelli, C., Binder, A., Braun, M., Kloft, M., & Muller, K. (2014). When brain and behavior disagree: Tackling systematic label noise in EEG data with machine learning. In 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014 [6782561] IEEE Computer Society. https://doi.org/10.1109/iww-BCI.2014.6782561

When brain and behavior disagree : Tackling systematic label noise in EEG data with machine learning. / Porbadnigk, Anne K.; Görnitz, Nico; Sannelli, Claudia; Binder, Alexander; Braun, Mikio; Kloft, Marius; Muller, Klaus.

2014 International Winter Workshop on Brain-Computer Interface, BCI 2014. IEEE Computer Society, 2014. 6782561.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Porbadnigk, AK, Görnitz, N, Sannelli, C, Binder, A, Braun, M, Kloft, M & Muller, K 2014, When brain and behavior disagree: Tackling systematic label noise in EEG data with machine learning. in 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014., 6782561, IEEE Computer Society, 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014, Gangwon, Korea, Republic of, 14/2/17. https://doi.org/10.1109/iww-BCI.2014.6782561
Porbadnigk AK, Görnitz N, Sannelli C, Binder A, Braun M, Kloft M et al. When brain and behavior disagree: Tackling systematic label noise in EEG data with machine learning. In 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014. IEEE Computer Society. 2014. 6782561 https://doi.org/10.1109/iww-BCI.2014.6782561
Porbadnigk, Anne K. ; Görnitz, Nico ; Sannelli, Claudia ; Binder, Alexander ; Braun, Mikio ; Kloft, Marius ; Muller, Klaus. / When brain and behavior disagree : Tackling systematic label noise in EEG data with machine learning. 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014. IEEE Computer Society, 2014.
@inproceedings{495a58e0caf94032822e9cc30eca13ef,
title = "When brain and behavior disagree: Tackling systematic label noise in EEG data with machine learning",
abstract = "Conventionally, neuroscientific data is analyzed based on the behavioral response of the participant. This approach assumes that behavioral errors of participants are in line with the neural processing. However, this may not be the case, in particular in experiments with time pressure or studies investigating the threshold of perception. In these cases, the error distribution deviates from uniformity due to the heteroscedastic nature of the underlying experimental set-up. This problem of systematic and structured (non-uniform) label noise is ignored when analysis are based on behavioral data, as is being done typically. Thus, we run the risk to arrive at wrong conclusions in our analysis. This paper proposes a remedy to handle this crucial problem: we present a novel approach for a) measuring label noise and b) removing structured label noise. We show its usefulness for an EEG data set recorded during a standard d2 test for visual attention.",
keywords = "Applied Cognitive Neuroscience, EEG, Label Noise, Machine Learning, Unsupervised Learning",
author = "Porbadnigk, {Anne K.} and Nico G{\"o}rnitz and Claudia Sannelli and Alexander Binder and Mikio Braun and Marius Kloft and Klaus Muller",
year = "2014",
month = "1",
day = "1",
doi = "10.1109/iww-BCI.2014.6782561",
language = "English",
booktitle = "2014 International Winter Workshop on Brain-Computer Interface, BCI 2014",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - When brain and behavior disagree

T2 - Tackling systematic label noise in EEG data with machine learning

AU - Porbadnigk, Anne K.

AU - Görnitz, Nico

AU - Sannelli, Claudia

AU - Binder, Alexander

AU - Braun, Mikio

AU - Kloft, Marius

AU - Muller, Klaus

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Conventionally, neuroscientific data is analyzed based on the behavioral response of the participant. This approach assumes that behavioral errors of participants are in line with the neural processing. However, this may not be the case, in particular in experiments with time pressure or studies investigating the threshold of perception. In these cases, the error distribution deviates from uniformity due to the heteroscedastic nature of the underlying experimental set-up. This problem of systematic and structured (non-uniform) label noise is ignored when analysis are based on behavioral data, as is being done typically. Thus, we run the risk to arrive at wrong conclusions in our analysis. This paper proposes a remedy to handle this crucial problem: we present a novel approach for a) measuring label noise and b) removing structured label noise. We show its usefulness for an EEG data set recorded during a standard d2 test for visual attention.

AB - Conventionally, neuroscientific data is analyzed based on the behavioral response of the participant. This approach assumes that behavioral errors of participants are in line with the neural processing. However, this may not be the case, in particular in experiments with time pressure or studies investigating the threshold of perception. In these cases, the error distribution deviates from uniformity due to the heteroscedastic nature of the underlying experimental set-up. This problem of systematic and structured (non-uniform) label noise is ignored when analysis are based on behavioral data, as is being done typically. Thus, we run the risk to arrive at wrong conclusions in our analysis. This paper proposes a remedy to handle this crucial problem: we present a novel approach for a) measuring label noise and b) removing structured label noise. We show its usefulness for an EEG data set recorded during a standard d2 test for visual attention.

KW - Applied Cognitive Neuroscience

KW - EEG

KW - Label Noise

KW - Machine Learning

KW - Unsupervised Learning

UR - http://www.scopus.com/inward/record.url?scp=84899419918&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84899419918&partnerID=8YFLogxK

U2 - 10.1109/iww-BCI.2014.6782561

DO - 10.1109/iww-BCI.2014.6782561

M3 - Conference contribution

AN - SCOPUS:84899419918

BT - 2014 International Winter Workshop on Brain-Computer Interface, BCI 2014

PB - IEEE Computer Society

ER -