仅更新9%参数即可高效适配脑电基础模型，适合资源受限的临床场景。
Parameter-Efficient Self-Supervised Adaptation for EEG-FM under Fixed Computational Budgets
- 仅更新9%参数，通过自监督微调对齐目标任务表征。
- 在三个临床数据集上，性能提升最高达20倍AUCPR，用20-50%数据即达峰值。
- 固定窗口数时，患者数量不影响性能，关键在于时间窗口多样性。
基于自监督学习预训练的脑电基础模型（EEG-FM）虽具备可迁移表征能力，但在跨不同临床数据集时泛化性仍受限。全量微调因计算开销过高，在资源受限的临床环境中不可行。本文研究参数高效的自监督适配方法，仅更新9%参数是否足以对齐目标任务表征。我们在两种具有不同预训练目标的先进模型（BIOT，对比学习；CBraMod，掩码重建）上评估，使用三个临床脑电数据集进行异常检测（TUAB）、事件分类（TUEV）和癫痫发作检测（CHB-MIT），涵盖分布内与分布外条件。结果表明，自监督适配相比线性探测持续提升性能，最高达20倍AUCPR。在固定计算预算下，仅需20%-50%的未标注数据即可达到最佳性能。关键发现：当总窗口数固定时，性能不随患者数量变化，说明性能仅依赖于整体时间窗口多样性。研究证明，参数高效适配可在极小计算开销和数据收集负担下有效部署EEG-FM。代码已开源：https://github.com/c3n-group/efficient-eeg-adapt
原文摘要 · Abstract (English)
EEG foundation models pretrained via self-supervised learning promise transferable representations, but their generalization remains limited, especially across diverse clinical datasets. Full fine-tuning is impractical for resource-constrained clinical settings due to high computational requirements. In this work, we investigate whether parameter-efficient self-supervised adaptation, updating only 9% of parameters suffices to align representations to target tasks. We evaluate our method on two state-of-the-art models with different pretraining objectives: BIOT (contrastive) and CBraMod (masked reconstruction), and evaluate on three clinical EEG datasets for abnormality detection (TUAB), event classification (TUEV), and seizure detection (CHB-MIT) under both in-distribution and out-of-distribution conditions. SSL adaptation yields consistent gains over linear probing, up to 20x AUCPR. Under a fixed compute budget, peak performance requires only 20--50% of available unlabeled data. Critically, when total window count is fixed, performance remains invariant to patient count, suggesting that performance is dependent on overall temporal window diversity only. Our findings demonstrate that parameter-efficient adaptation enables effective deployment of EEG Foundation models (EEG-FM) with minimal computational overhead and data collection burden. Code available at: https://github.com/c3n-group/efficient-eeg-adapt
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。