Dual Prototyping with Domain and Class Prototypes for Affective Brain-Computer Interface in Unseen Target Conditions

2024-11-27

arXiv.org

AI Summary

Cross-subject EEG-based emotion recognition faces two critical bottlenecks: reliance on target-domain data for model adaptation and susceptibility to label noise. Method: This paper proposes a pure source-domain transfer learning framework. It introduces a novel dual-prototype mechanism—comprising domain prototypes and class prototypes—to enable transferable representation learning without accessing any target-domain samples. The framework integrates feature disentanglement with domain-invariant modeling and incorporates a pairwise learning strategy to enhance robustness against label noise. Contribution/Results: Evaluated on SEED and SEED-IV benchmarks, the method achieves performance comparable to fully supervised transfer approaches while requiring zero target-domain data. It significantly improves model robustness, generalizability, and practical deployability in real-world cross-subject scenarios.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Representation Learning for Vision
• Application Domains: Humanities & Computational Social Science

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

EEG signals have emerged as a powerful tool in affective brain-computer interfaces, playing a crucial role in emotion recognition. However, current deep transfer learning-based methods for EEG recognition face challenges due to the reliance of both source and target data in model learning, which significantly affect model performance and generalization. To overcome this limitation, we propose a novel framework (PL-DCP) and introduce the concepts of feature disentanglement and prototype inference. The dual prototyping mechanism incorporates both domain and class prototypes: domain prototypes capture individual variations across subjects, while class prototypes represent the ideal class distributions within their respective domains. Importantly, the proposed PL-DCP framework operates exclusively with source data during training, meaning that target data remains completely unseen throughout the entire process. To address label noise, we employ a pairwise learning strategy that encodes proximity relationships between sample pairs, effectively reducing the influence of mislabeled data. Experimental validation on the SEED and SEED-IV datasets demonstrates that PL-DCP, despite not utilizing target data during training, achieves performance comparable to deep transfer learning methods that require both source and target data. This highlights the potential of PL-DCP as an effective and robust approach for EEG-based emotion recognition.

Problem

• Addresses EEG emotion recognition under unseen target conditions
• Reduces model dependency on source and target domains
• Mitigates label noise effects in emotion recognition

Innovation

• Pairwise Learning with Domain and Class Prototypes
• Feature disentanglement for EEG emotion recognition
• Reduces label noise effect via pairwise learning