ProDG: A proxy-domain-guiding strategy for multi-source-free domain adaptation in EEG emotion recognition

• Published in: Knowledge-based systems Vol. 329; p. 114318
• Main Authors: Zhou, Bingtao, Xiang, Mian, Ning, Qian
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 04.11.2025
• Subjects: Domain adaptation; EEG emotion recognition; Multi-source-free domain adaptation; Proxy; Domain adaptation; EEG emotion recognition; Proxy; Multi-source-free domain adaptation
• ISSN: 0950-7051
• DOI: 10.1016/j.knosys.2025.114318

•Research highlight 1 Aiming to improve the performance of multi-source-free domain adaptation in privacy-preserving EEG emotion recognition, we propose and theoretically validate that high-confidence predictions from source models can construct a proxy domain approximating the target domain, providing a theoretical basis for complementary source guidance without accessing raw source data.•Research highlight 2 To operationalize this theory, we propose a Proxy-Domain-Guiding (ProDG) strategy consist of two modules: Proxy Mutual Information Alignment (PrMI) constructs a proxy domain through aggregated high-confidence source predictions, then aligns source models via mutual information maximization; Proxy Pseudo-Label Alignment (PrPL) refines pseudo-labels with cross-source confidence validation to enhance supervision.•Research highlight 3 Evaluations on EEG benchmarks SEED, SEED-IV and DEAP demonstrate ProDG’s superior privacy-preserving performance. [Display omitted] Subject-independent Electroencephalogram (EEG) emotion recognition has underperformed due to significant disparities among subjects. Domain adaptation (DA) is a common solution, but traditional methods require access to target domain data, raising privacy concerns. Source-free domain adaptation offers a viable solution, however, researches on it remains unexplored. Moreover, existing methods overlooked the complementary information across source domains. To overcome this challenge, we focus on exploring the inter-domain complementarity. Our core insight is that higher-confidence predictions from source models indicate regions closer to the target domain’s distribution. Based upon, we propose Proxy-Domain-Guiding (ProDG) strategy, which pioneers confidence-guidance to achieve privacy-preserving recognition. First, we propose a Proxy Guiding theory validating that predictions of source models with higher confidence exhibit closer distributional proximity to the target domain. Then, we propose two modules: Proxy Mutual Information Alignment (PrMI) constructs a proxy domain by aggregating high-confidence predictions from source models, approximating the target-overlapping region, then each source model is aligned with proxy domain via mutual information maximization; Proxy Pseudo-Label Alignment (PrPL) refines clustering-based pseudo-labels using cross source confidence evaluation, enhancing supervised loss quality. The whole training process utilize only the source domain model and target data, with source data being inaccessible, ensuring privacy-preserving. Our method attains state-of-the-art accuracy on DEAP(65.3 %), SEED (85.9 %) and SEED-IV (70.4 %), surpassing privacy-preserving methods by a large margin and rivaling non-privacy-preserving approaches. ProDG validates the efficacy of confidence-based proxy guiding in multi-source-free domain adaptation. This work was conducted at the College of Electronics and Information Engineering, Sichuan University in May 2025.