Scholarlink

Machine learning is essential to the development of personalized medicine, brain computer interfaces (BCIs), classification and prediction, and the detection and elimination of artifacts in EEG signal data, among other applications. This work, in order to differentiate between target and non-target rapid serial visual presentation (RSVP) experimental conditions predicts the spatiotemporal patterns of entire trial types. We developed an optimized pipeline to preprocess EEG time-series data in a way that maximizes the relevance of event-related potentials (ERPs). We then utilized the machine learning techniques with the open-source EEG software, namely the MNE-Python tools (library), using the performance criteria, area under the receiver operating characteristic curve (ROC-AUC) with fivefold cross-validation to predict the trial types.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Improving Machine Learning-Based Activity Type Prediction from Time-Series EEG Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Syed Muhammad Raza Abidi,
  • Tomas Emmanuel Ward,
  • David C. Henshall,
  • Gabriel-Miro Muntean

摘要

Machine learning is essential to the development of personalized medicine, brain computer interfaces (BCIs), classification and prediction, and the detection and elimination of artifacts in EEG signal data, among other applications. This work, in order to differentiate between target and non-target rapid serial visual presentation (RSVP) experimental conditions predicts the spatiotemporal patterns of entire trial types. We developed an optimized pipeline to preprocess EEG time-series data in a way that maximizes the relevance of event-related potentials (ERPs). We then utilized the machine learning techniques with the open-source EEG software, namely the MNE-Python tools (library), using the performance criteria, area under the receiver operating characteristic curve (ROC-AUC) with fivefold cross-validation to predict the trial types.