脑机接口 / BCI - arXivDaily 专题

2608.11656 2026-08-13 cs.LG 新提交 91%

Continuous-Latent Predictive Modeling with Semantic Alignment for EEG-Language Foundation Models

面向EEG-语言基础模型的语义对齐连续隐式预测建模

Myeong-Ju Cho, Hye-Bin Shin, Seo-Hyun Lee, Seong-Whan Lee

专题命中 EEG解码：EEG(title,title_cn)；neural decoding(abstract)；分类 cs.LG

AI总结本文提出BLPM模型，通过CELP编码器与MQSD模块对齐EEG语义，解决EEG基础模型预训练的关键挑战，在多基准任务中实现良好泛化。

Comments 19 pages, 3 figures; supplementary material included

2608.12000 2026-08-13 q-bio.NC cs.LG eess.SP 新提交 83%

Beyond Local Power: Functional Connectivity Analysis for Subject-Independent Learning Style Recognition

超越局部能力：用于被试独立学习风格识别的功能连接分析

Wiga Maulana Baihaqi, Indriana Hidayah, Sri Kusrohmaniah, Noor Akhmad Setiawan

专题命中 EEG解码：EEG(summary_cn,abstract)；分类 q-bio.NC、eess.SP、cs.LG

AI总结该研究提出基于EEG和PLV连接性的方法，识别被试独立学习风格，在VV维度获70.00%准确率，AR维度仅55.56%，指出需自适应特征变换缩小跨被试泛化差距。

Comments 7 pages, 5 figures. Accepted for publication at 4th IEEE International Conference on Artificial Intelligence and Mechatronics Systems 2026

2608.12072 2026-08-13 stat.ME q-bio.NC 新提交 80%

Testing the limits of past-adapted explanations by post-endpoint randomisation: anticipatory EEG as a worked case

通过端点后随机化测试适配过去的解释的极限：以预期脑电图为例

George Sopasakis, Alexandros Sopasakis

专题命中 EEG解码：EEG(title,abstract)；分类 q-bio.NC

AI总结本研究提出Level II-A框架，以合成脑电图数据为案例，通过端点后随机化测试适配过去的解释的极限，将“过去解释”转化为可检验的量级限定主张。

Comments Perspective article. 90 pages, 6 figures, 15 tables; includes Supplementary Information. No human EEG data are analysed. Synthetic benchmark and reproducible software are available at https://github.com/geosop/LEVEL-IIA and https://doi.org/10.5281/zenodo.21887583

2510.11917 2026-08-13 cs.LG 版本更新 79%

Variational Mixture of Graph Neural Experts for Alzheimer's Disease Recognition across Frequency Bands in EEG Brain Networks

用于跨脑电图脑网络频段的阿尔茨海默病识别的变分图神经专家混合模型

Jun-En Ding, Anna Zilverstand, Shihao Yang, Albert Chih-Chieh Yang, Feng Liu

机构 * Department of Industrial and Systems Engineering, Rutgers University（鲁特大学工业与系统工程系）； Department of Psychiatry and Behavioral Sciences, University of Minnesota（明尼苏达大学精神病学与行为科学系）； Institute of Brain Science, College of Medicine, National Yang-Ming Chiao Tung University（国家阳明交通大学医学院脑科学研究所）

专题命中 EEG解码：EEG(title,abstract)；分类 cs.LG

AI总结针对痴呆症脑电图诊断难题，提出VMoGE框架，结合多频段脑电图分析、变分图神经网络和专家混合架构。实验显示其性能强大，还具有临床转化价值，能关联疾病指标、揭示频段变化，为痴呆症诊断提供新方法和依据。