📄 中文摘要
基于脑电图（EEG）的情感识别在跨异构数据集转移时面临严重的性能下降，这主要源于生理变异、实验范式差异和设备不一致性。现有的领域对抗方法主要强调全局边际对齐，往往忽视类条件不匹配和决策边界扭曲，从而限制了跨语料库的泛化能力。提出了一种统一的原型驱动对抗对齐（PAA）框架，用于跨语料库的 EEG 情感识别。该框架逐步实现了三种配置：PAA-L，进行原型引导的局部类条件对齐；PAA-C，进一步增强了对抗对齐的效果。
边界感知原型驱动对抗对齐用于跨语料库 EEG 情感识别
📄 English Summary
Boundary-aware Prototype-driven Adversarial Alignment for Cross-Corpus EEG Emotion Recognition
Electroencephalography (EEG)-based emotion recognition experiences significant performance degradation when models are transferred across heterogeneous datasets due to physiological variability, differences in experimental paradigms, and inconsistencies in devices. Existing domain adversarial methods primarily focus on enforcing global marginal alignment, often neglecting class-conditional mismatches and decision boundary distortions, which limits cross-corpus generalization. A unified Prototype-driven Adversarial Alignment (PAA) framework is proposed for cross-corpus EEG emotion recognition. The framework is progressively instantiated in three configurations: PAA-L, which performs prototype-guided local class-conditional alignment; PAA-C, which further enhances the effectiveness of adversarial alignment.