Recent research:
This body of work advances neuroengineering and wearable AI by integrating machine learning with neurophysiological signals to model, predict, and enhance cognitive and communicative functions. The author combines EEG-based brain data, advanced neural architectures, and rigorous feature selection to (1) characterize individual differences in visual working memory through spectral EEG patterns and spatiotemporal analysis, enabling subject-specific task customization and potential neurofeedback via brain-computer interfaces; (2) improve ERP-based brain–computer interface performance by post-processing recurrent neural network outputs with a Post-Recurrent Module that enhances decision accuracy and model interpretability; and (3) evaluate the impact of mobile device constraints on AI-driven emotion recognition, identifying architecture choices that balance accuracy with computational efficiency. A complementary approach targets P300 detection by selecting informative neurons in recurrent networks to form a post-validation ensemble, boosting sample-level temporal structure detection and enabling real-time ERP monitoring. Overall, the research contributes methodological advances in signal processing, ML architecture design, and practical deployment considerations for neurotechnology applications across rehabilitation, human–machine interaction, and mobile sensing; with implications for personalized neurofeedback, robust BCIs, and accessible mobile AI tools.