Exploration of Person-Independent BCIs for Internal and External Attention-Detection in ...
summary
16-channel EEG plus binocular eye tracking record 15–20s labeled segments of internal vs external attention from 14 participants and test LDA, random forest, a vanilla NN, and convolutional/shallow NNs across 1–13s windows. EEG-only reaches ~80% person-dependent accuracy, eye tracking 67%, combined ~82% (not significant). A shallow neural net on 4-second raw EEG windows suits real-time person-independent use, showing 88% peak accuracy on a selected training set and an average 60% person-independent rate (95% CI 54–65%).
AI-written summary · about
related talks
01h 05mUsing EEG and Azure Machine Learning to Perform Lie Detection
31mJennifer Marsman, Principal Developer Evangelist, Microsoft
24mControlling Apps with Your Mind and AI
24mThe Brainduino Project, Brain-Computer Interfaces & AI in the Future
32mThe Missing Layer in the AI Stack: Cognitive Performance for the AI Era
23mParallel Eyes: Exploring Human Capability and Behaviors with Paralleled First Person View Sharing
06mCES Interview
09mQuantitative Evaluation of Machine Learning Explanations: A Human-Grounded Benchmark