Audiovisual stimulus classification (MNE sample dataset)

Name: audiovisual_stimulus

Category: sensory

Dataset: Mne2013SampleEeg

Objective: Multiclass classification

Split: Stratified trial-level split (single subject)

Usage

neuralbench eeg audiovisual_stimulus

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: Mne2013SampleEeg
    split:
      name: SklearnSplit
      split_by: _index
      valid_split_ratio: 0.2
      test_split_ratio: 0.2
      valid_random_state: 33
      test_random_state: 33
      stratify_by: description
  neuro.baseline: [0.0, 0.2]
  channel_positions.layout_or_montage_name: null  # Use original channel positions
  target:
    =replace=: true
    name: LabelEncoder
    event_types: Stimulus
    event_field: description
    return_one_hot: true
    aggregation: trigger
  trigger_event_type: Stimulus
  start: -0.2
  duration: 1.0
  summary_columns: [description]
brain_model_output_size: &brain_model_output_size 4
trainer_config.monitor: val/bal_acc
trainer_config.mode: max
loss:
  name: CrossEntropyLoss
  kwargs:
    label_smoothing: 0.1
metrics: !!python/object/apply:neuralbench.defaults.metrics.get_classification_metric_configs
  - *brain_model_output_size

Description

This task involves classifying audiovisual stimuli from EEG recordings into four categories: left ear auditory stimulus, right ear auditory stimulus, left visual field stimulus, or right visual field stimulus. The task uses the MNE-Python sample dataset [MNESample], which is a small, publicly available dataset designed for testing and demonstration purposes. We use it here as an example of a low data regime task.

Note

This is a single-subject dataset intended primarily for sanity-checking and integration testing. Because only one subject is available, the train/validation/test split is performed at the trial level (split_by: _index).

References

[MNESample]

1. Gramfort, M. Luessi, E. Larson, D. Engemann, D. Strohmeier, C. Brodbeck, L. Parkkonen, M. Hämäläinen. MNE software for processing MEG and EEG data. NeuroImage, 86, 446-460, 2014. https://mne.tools/stable/documentation/datasets.html#sample