固定计算预算下EEG-FM的参数高效自监督适配
Parameter-Efficient Self-Supervised Adaptation for EEG-FM under Fixed Computational Budgets
- University of Tübingen（蒂宾根大学）
- Hertie Institute for AI in Brain Health (Hertie AI)（赫蒂脑健康人工智能研究所）
- University Clinic Tübingen（蒂宾根大学医院）
- Hertie Institute for Clinical Brain Research（赫蒂临床脑研究所）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本研究提出参数高效自监督适配方法，仅更新9%参数即可适配EEG-FM，在三个临床EEG数据集上较线性探测获最高20倍AUCPR，固定计算预算下仅需20%-50%未标记数据，可降低部署负担。
AI中文摘要：
通过自监督学习预训练的EEG基础模型（EEG-FM）有望生成可迁移的表征，但其泛化能力仍有限，尤其在不同临床数据集间的表现不足。全量微调因计算资源需求高，在资源受限的临床场景中不具备可行性。本研究探究参数高效的自监督适配是否可行：仅更新9%的参数即可使表征与目标任务对齐。我们在两个采用不同预训练目标的最先进模型上评估所提方法：BIOT（对比式）和CBraMod（掩码重构式），并在三个临床EEG数据集上开展评估，涵盖异常检测（TUAB）、事件分类（TUEV）和癫痫发作检测（CHB-MIT），评估条件包括分布内与分布外两种情况。自监督适配（SSL adaptation）相较于线性探测方法可获得稳定提升，AUCPR最高可达20倍。在固定计算预算下，达到峰值性能仅需20%-50%的可用未标记数据。关键发现是，当总窗口数固定时，性能与患者数量无关，表明性能仅取决于整体时间窗口的多样性。本研究结果表明，参数高效适配可实现EEG-FM的有效部署，且计算开销与数据采集负担均极小。代码可获取于此https URL。
英文摘要：
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
