针对少量脑电数据，提出自适应分布对齐方法提升跨数据集识别准确率。
Adaptive Split-MMD Training for Small-Sample Cross-Dataset P300 EEG Classification
- 设计自适应分块最大均值差异训练，动态调整源目标域权重与归一化策略。
- 在10样本/人目标数据下，准确率66%、AUC达74%，优于传统迁移方法。
- 适合小样本脑电分类任务，尤其适用于资源受限的神经工程应用。
当仅有少量标注试次时，从脑电中检测单次事件相关电位（P300）极具挑战性。通过迁移学习利用大规模源数据增强小规模目标数据时，会面临跨数据集分布偏移问题。本文在严格小样本设定下（目标：每被试10次试次；源：每被试80次试次），研究两个公开视觉奇异性诱发脑电数据集之间的迁移，仅使用五个共享电极（Fz, Pz, P3, P4, Oz）。提出自适应分块最大均值差异训练（AS-MMD），结合三项创新：(i) 基于源/目标样本量比平方根的温启动加权损失，(ii) 共享仿射参数但保留各域独立运行统计的分块归一化（Split-BN），(iii) 无需参数的基于中位数带宽启发式的对数层径向基函数核最大均值差异（RBF-MMD）。该方法部署于脑电转换器（EEG Conformer），具备模型无关性且推理阶段无需修改。在双向迁移中，其性能均优于仅用目标数据和合并训练（主动视觉奇异性：准确率/0.66，AUC 0.74；ERP CORE P3：0.61，AUC 0.65），经校正配对t检验，合并训练优势显著。消融实验表明三者贡献均关键。
原文摘要 · Abstract (English)
Detecting single-trial P300 from EEG is difficult when only a few labeled trials are available. When attempting to boost a small target set with a large source dataset through transfer learning, cross-dataset shift arises. To address this challenge, we study transfer between two public visual-oddball ERP datasets using five shared electrodes (Fz, Pz, P3, P4, Oz) under a strict small-sample regime (target: 10 trials/subject; source: 80 trials/subject). We introduce Adaptive Split Maximum Mean Discrepancy Training (AS-MMD), which combines (i) a target-weighted loss with warm-up tied to the square root of the source/target size ratio, (ii) Split Batch Normalization (Split-BN) with shared affine parameters and per-domain running statistics, and (iii) a parameter-free logit-level Radial Basis Function kernel Maximum Mean Discrepancy (RBF-MMD) term using the median-bandwidth heuristic. Implemented on an EEG Conformer, AS-MMD is backbone-agnostic and leaves the inference-time model unchanged. Across both transfer directions, it outperforms target-only and pooled training (Active Visual Oddball: accuracy/AUC 0.66/0.74; ERP CORE P3: 0.61/0.65), with gains over pooling significant under corrected paired t-tests. Ablations attribute improvements to all three components.
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。