Use Cases Code Examples

Use Cases Overview

Labels

• (multi-label) categorical/continuous (classification/regression)
• Per-trial/per-timestep
• Multiple Targets (speed precision)
• Conflicting Labels from multiple raters
• Non-EEG channels (should be fine, may get resampled to same resolution)

Temporal Aspects:

• Single (temporal) prediction training / multiple (temporal) prediction training
  • Especially important regarding correct (causal) evaluation
• Variable trial length/fixed trial length

Data Aspects:

• Within-subject/across-subject
  • with and without pretraining

Optional:

• Handling missing data
• Confounder that should be ignored (domain-adversarial)

Example codes

Categorical per-trial single-prediction fixed length within-subject

Realistic example: https://github.com/braindecode/braindecode/blob/master/examples/bcic_iv_2a.py (with Experiment class) Simple example: https://github.com/braindecode/braindecode/blob/master/examples/plot_trial_decoding.py