HEDN: A Hard-Easy Dual Network with Source Reliability Assessment for Cross-Subject EEG Emotion Recognition

Cross-subject electroencephalography (EEG) emotion recognition is essential for real-time monitoring of cognitive and affective states in brain-computer interface (BCI) and wearable health applications, but substantial inter-subject variability poses a major challenge. Multi-Source Domain Adaptation (MSDA) offers a potential solution, but existing MSDA frameworks typically assume equal source quality, leading to negative transfer from low-reliability domains and prohibitive computational overhead due to multi-branch model designs. To address these limitations, we propose the Hard-Easy Dual Network (HEDN), a lightweight reliability-aware MSDA framework. HEDN introduces a novel Source Reliability Assessment (SRA) mechanism that dynamically evaluates the structural integrity of each source domain during training. Based on this assessment, sources are routed to two specialized branches: an Easy Network that exploits high-quality sources to construct fine-grained, structure-aware proto types for reliable pseudo-label generation, and a Hard Net work that improves the discriminability of low-reliability sources while regularizing source-target alignment. Furthermore, a cross-network consistency loss aligns pre dictions between branches to preserve semantic coherence. Extensive experiments conducted on SEED, SEED IV, and DEAP datasets demonstrate that HEDN achieves highly competitive performance compared with state-of the-art methods under cross-subject evaluation protocols while reducing adaptation complexity. The source code is available at https://github.com/qwangwl/HEDN.