Mental Workload Classification

Name: mental_workload
Category: brain-computer interfacing
Dataset: Hinss2022
Objective: Multiclass classification
Split: Leave-subjects-out

Usage

neuralbench eeg mental_workload
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: Hinss2023Open
    filter_task:
      name: QueryEvents
      query: "task in ['MATBeasy', 'MATBmed', 'MATBdiff']"
    split:
      name: SklearnSplit
      split_by: subject
      valid_split_ratio: 0.2
      test_split_ratio: 0.2
      valid_random_state: 33
      test_random_state: 33
  target:
    =replace=: true
    name: LabelEncoder
    event_types: Eeg
    event_field: task
    return_one_hot: true
    aggregation: trigger
  trigger_event_type: Eeg
  start: 0.0
  duration: 5.0
  stride: 5.0
  stride_drop_incomplete: true
  summary_columns: [task]
brain_model_output_size: &brain_model_output_size 3
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 mental workload levels from EEG recordings. The Hinss2022 dataset [Hinss2022] contains recordings from 29 participants performing 4 cognitive tasks designed to elicit distinct mental states: Psychomotor Vigilance Task (PVT), Flanker task, N-back tasks (zero-back, one-back, two-back), and Multi-Attribute Task Battery at different difficulty levels (easy, medium, difficult).

Here, we focus on the MATB task data, and use the different MATB levels (easy, medium, difficult) as the target labels to represent varying degrees of working memory load.

Additional Datasets

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

  • Hinss2021 – 15 subjects, 4 classes (resting state, easy, medium, difficult)

  • Jao2021 (BNCI2022_001) – 13 subjects, drone piloting at varying difficulty levels

To run with an alternate dataset:

neuralbench eeg mental_workload --datasets hinss2021

References

[Hinss2022]

Hinss, Marcel F., et al. “Open multi-session and multi-task EEG cognitive Dataset for passive brain-computer Interface Applications.” Scientific Data 10.1 (2023): 85.