P3 classification

Name: p3
Category: brain-computer interfacing
Dataset: Schreuder2010New (BNCI2015_009)
Objective: Binary classification
Split: Leave-subjects-out

Usage

neuralbench eeg p3
Show config.yaml
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

data:
  study:
    source:
      name: Schreuder2010New
    split:
      name: SklearnSplit
      split_by: subject
      valid_split_ratio: 0.2
      test_split_ratio: 0.2
      valid_random_state: 33
      test_random_state: 33
  neuro.baseline: [0.0, 0.2]
  target:
    =replace=: true
    name: LabelEncoder
    event_types: Stimulus
    event_field: code
    return_one_hot: true
    aggregation: trigger
  trigger_event_type: Stimulus
  start: -0.2
  duration: 1.0
  summary_columns: [code]
compute_class_weights: true
brain_model_output_size: &brain_model_output_size 2
trainer_config.monitor: val/bal_acc
trainer_config.mode: max
trainer_config.patience: 10
loss:
  name: CrossEntropyLoss
  kwargs:
    label_smoothing: 0.1
metrics: !!python/object/apply:neuralbench.defaults.metrics.get_classification_metric_configs
  - *brain_model_output_size

Description

The P3 classification task involves identifying the presence of P3 event-related potentials (ERPs) in EEG recordings. In this task, we use the Schreuder2010 dataset [Schreuder2010], which contains 20-channel EEG data from 21 healthy participants attending to spatial auditory cues in a multi-class AMUSE ERP paradigm (Target vs. NonTarget).

Additional Datasets

The following additional datasets from MOABB can also be used with this task:

  • Acqualagna2013 (BNCI2015_010) – 12 subjects, RSVP P300 speller

  • Arico2013 (BNCI2014_009) – 10 subjects, row-column + GeoSpell P300

  • Cattan2019Vr – 21 subjects, VR P300 speller

  • Guger2009 (BNCI2015_003) – 10 subjects, auditory P300

  • Haufe2011 (BNCI2016_002) – 15 subjects, emergency braking ERP

  • Hoffmann2008 (EPFLP300) – 8 subjects

  • Huebner2016 – visual P300 matrix speller (mix-mode)

  • Huebner2017 – 13 subjects, visual P300 matrix speller

  • Kappenman2021P3 (ErpCore2021_P3) – 40 subjects, visual oddball P300

  • Kojima2024A – 11 subjects

  • Kojima2024B – 15 subjects

  • Korczowski2014A (BI2014a) – 64 subjects, Brain Invaders P300

  • Korczowski2014B (BI2014b) – 38 subjects

  • Korczowski2015A (BI2015a) – 43 subjects

  • Korczowski2015B (BI2015b) – 44 subjects

  • Lee2019Erp – 54 subjects, P3 speller with face-overlay stimuli

  • Riccio2013 (BNCI2014_008) – 8 subjects (ALS patients)

  • Romani2025 – 22 subjects, BrainForm serious-game P300

  • Schaeff2012 (BNCI2015_007) – 16 subjects, motion-onset VEP P300

  • Sosulski2019 – 13 subjects, auditory oddball P300

  • Treder2011 (BNCI2015_008) – 13 subjects, center-speller P300

  • Treder2014 (BNCI2015_006) – 11 subjects, auditory BCI

  • VanVeen2019 (BI2012) – 25 subjects, Brain Invaders P300

To run with an alternate dataset:

neuralbench eeg p3 --dataset kappenman2021p3

References

[Schreuder2010]

Schreuder, Martijn, Benjamin Blankertz, and Michael Tangermann. “A New Auditory Multi-Class Brain-Computer Interface Paradigm: Spatial Hearing as an Informative Cue.” PLoS ONE 5.4 (2010): e9813.