A lightweight EEG-based cognitive state recognition method integrating vision transformer and genetic algorithms for multi-tasking product design

Le résumé fourni par la source

During complex product design tasks, designers’ cognitive states undergo continuous and dynamic changes aligned with evolving design intentions. Accurately and efficiently identifying cognitive states is crucial during complex design tasks since it enables AI-driven design systems understand designer’s intention in real time and facilitates proactive human–AI collaboration. To achieve it, electroencephalography (EEG) data is widely used but EEG-based cognitive state recognition task still faces the challenges of long-time EEG device settings and insufficient generalisation of channel-reduced models in complex product design scenarios. To address these issues, this study proposes a lightweight EEG-based cognitive state recognition framework integrating Vision Transformer (ViT) and genetic algorithms (GA). First, a rapid EEG channel selection method is introduced to generate an efficient channel configuration scheme. Second, a detailed description of the experimental procedure for active knowledge recommendation based on EEG is provided, along with the presentation and analysis of results from various complex design tasks. Third, comprehensive validation through experiments and case studies demonstrates that the high-precision channel configuration method proposed in this study significantly improves the deployment efficiency of EEG devices while maintaining recognition accuracy, offering an efficient cognitive state recognition solution for the field of complex product design.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.