Support vector machines to detect physiological patterns for EEG and EMG-based human-computer interaction: a review.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Human-Computer Interaction

Background:

• Support Vector Machines (SVMs) are prevalent classifiers in Human-Computer Interaction (HCI) for physiological pattern detection.
• Their effectiveness stems from versatility, robustness, and accessible toolboxes.
• However, inconsistent reporting of SVM implementation details hinders result reproducibility in scientific literature.

Purpose of the Study:

• To provide a comprehensive critical review of SVM applications in analyzing electroencephalography (EEG) and electromyography (EMG) data for HCI.
• To address the lack of detailed methodology in existing studies.
• To facilitate optimized classification and accurate reporting of results.

Main Methods:

• Review of existing literature focusing on SVM usage for EEG and EMG in HCI.
• Outline of fundamental SVM theory.
• Compilation of implementation details and parameter selection from reviewed papers into tables.
• Statistical analysis of SVM application trends.

Main Results:

• Identification of common SVM implementations and their parameters in EEG/EMG-based HCI research.
• Presentation of statistics illustrating the prevalence of SVM in the field.
• Critical comparison of SVM performance against alternative classification methods.

Conclusions:

• SVMs are highly suitable for brain and muscle pattern detection in HCI applications.
• Standardized reporting of SVM parameters and methodology is crucial for reproducibility.
• Further research should focus on optimizing SVM configurations for specific HCI tasks.