核心矩阵协议 · 迁移到新被试

心算与静息（EEGMAT）

下面这些分数是怎么来的——队列、数据划分、电极、时间窗、每种方法允许学什么、随机猜测能得多少分——都按已发布的协议文件给出。只在本协议内部比较分数。

协议步骤、局限说明、方法备注与数据集署名来自发布数据本身，保持英文原文——它们随数据一起被审核，翻译会让网页与可下载文件不再一致。

协议概览

任务: Serial subtraction versus resting EEG
评测方式: 迁移到新被试
数据集: EEGMAT
被试: 36
数据量: 2,160 epochs · 5 participant-disjoint folds
输入: 19 channels · 2-second windows
随机水平: 50.0%
指标: 平衡准确率（主指标）· 宏平均 F1（次指标）

结果

在本协议下运行过的每一种方法，以及已发布结果文件中记录的分数。只在这张表内纵向比较：其他协议在队列、电极、时间窗或随机水平上有所不同。

Spectral ridge

56.8% (54.6%–58.9%)

LaBraM

64.6% (60.5%–68.6%)

CBraMod

62.3% (58.1%–66.5%)

EEGNet

67.6% (62.8%–72.5%)

平衡准确率，本协议下的每一种方法。 点为估计值，横线为描述性 95% 区间，虚线为随机水平。

方法	训练方式	平衡准确率	宏平均 F1	通道	被试	评分耗时
Spectral ridge经典方法	Supervised fit	56.8%54.6%–58.9%	0.558Mean across held-out participants	19	36	0.6 s
LaBraM基础模型	Frozen encoder + ridge head	64.6%60.5%–68.6%	0.620Mean across held-out participants	19	36	2.1 s
CBraMod基础模型	Frozen encoder + ridge head	62.3%58.1%–66.5%	0.602Mean across held-out participants	19	36	6.1 s
EEGNet轻量模型	Scratch · 20 epochs	67.6%62.8%–72.5%仅单个随机种子——为已运行种子中最高的一次（见「稳定性」）	0.637Mean across held-out participants	19	36	117.7 s

评分耗时包括拟合与预测，可能含加速器等待时间，不含数据准备。它只对该配置有效，不是硬件基准。

未在本协议下运行： CSP+LDA, ShallowFBCSPNet, Deep4Net, Standard CCA, Temporal ridge。这里缺少的方法只是没有在本协议上运行——不是失败。

谨慎解读

Thirty nonoverlapping 2-second windows from each of two conditions per person. Published signals were already cleaned with ICA; task performance groups are not evaluated.

协议步骤

5 participant-disjoint folds. All recordings from a person stay together. Each person contributes to the held-out predictions once.
First 60 s of each recording; 30 contiguous nonoverlapping 2 s windows; exclude A2-A1 ear-difference and ECG; EDF physical volts converted to microvolts; subtract each channel's window mean; no rejection, filtering, resampling, or learned preprocessing.
One fixed seed (20260919); no early stopping or test-based tuning. EEGNet trains for 20 epochs per fold. Frozen encoders use training-only standardized ridge heads (alpha 100).
Labels: pre-task-rest, mental-arithmetic.
Uniform-guessing reference: 50.00%. Scores weight participants equally. Intervals describe participant variation; cross-validation training sets overlap.
The benchmark detects condition (rest versus serial subtraction), not the good/bad count-quality participant grouping.
Only the first documented 60 seconds is retained even though EDF containers are longer.
The source README reports prior ICA artifact removal, so these are not untouched acquisition signals.
Open Data Commons Attribution License v1.0 applies; retain PhysioNet attribution.
EEGNet three-seed mean 67.52%; sample SD 0.14 percentage points; range 67.36–67.64%. Main table retains the original fixed seed; this is not a confidence interval.

稳定性

EEGNet three-seed mean 67.52%; sample SD 0.14 percentage points; range 67.36–67.64%. Main table retains the original fixed seed; this is not a confidence interval.

EEGNet 各随机种子的结果： 67.64% · 67.55% · 67.36%；均值 67.52%

Same participants, folds, preprocessing and 20-epoch budget. Three seeds measure initialization variability, not population uncertainty. Main table retains its preselected seed; no best-seed selection.

是否出现在预训练数据中

Unknown unless explicitly documented; no unseen-pretraining claim.

方法说明

One fixed configuration. Foundation encoders remain frozen; small networks train from scratch. These scores do not establish optimal fine-tuned performance. No individual predictions or participant-level results are distributed.
One fixed configuration. Foundation encoders remain frozen; small networks train from scratch. These scores do not establish optimal fine-tuned performance. No individual predictions or participant-level results are distributed. EEGNet three-seed mean 67.52%; sample SD 0.14 percentage points; range 67.36–67.64%. Main table retains the original fixed seed; this is not a confidence interval.

模型条款

Spectral ridge：Trained from scratch / deterministic reference; no third-party pretrained weights. Braindecode BSD-3-Clause; MNE/scikit-learn BSD where used.
LaBraM：Code/repository: MIT · Checkpoint: committed in that repository; no separate weight terms
CBraMod：Code: MIT · Weights: Apache-2.0 (official model card)
EEGNet：Trained from scratch / deterministic reference; no third-party pretrained weights. Braindecode BSD-3-Clause; MNE/scikit-learn BSD where used.

来源与许可

EEGMAT

署名 Igor Zyma, Ivan Seleznov, Anton Popov, Mariia Chernykh, Oleksii Shpenkov · EEG During Mental Arithmetic Tasks 1.0.0, PhysioNet, doi:10.13026/C2JQ1P. Study: Zyma et al. (2019), doi:10.3390/data4010014. PhysioNet platform: Pollard et al. (2026), doi:10.1038/s44360-026-00096-z.

许可 Open Data Commons Attribution License 1.0 ↗

数据集记录 ↗ · 这个数据集上的全部结果 →

BCI Report 不转发任何记录。这些是 BCI Report 在上述许可下计算的聚合测量；数据归属于署名中的作者。

许可范围 Personal noncommercial research; aggregate results only

隐私这里只发布队列级聚合结果：不发布任何记录、被试编号或逐人分数。完整的审查说明在协议 JSON 中 ↓

公开数据登记说明 署名许可与该研究专属的伦理批准、同意书都有记录；可以使用的前提是只输出聚合结果，并排除被试信息字段。Attribution license and study-specific approval/consent are documented; eligibility assumes aggregate-only output and exclusion of subject-info fields.

权利审查于 2026-09-20 · 依据 physionet.org ↗ · www.mdpi.com ↗ · opendatacommons.org ↗

复现记录

协议 ID: parallel-fixed-subject-folds-v1/physionet-eegmat-1.0.0
数据集版本: PhysioNet EEG During Mental Arithmetic Tasks 1.0.0 · 1.0.0
硬件: Apple M5 / MPS
评分阶段总耗时: 126.5 s · 可能含加速器等待
审计记录 SHA-256: 2d0e4cec78c8d6e5bfc9f5390ba161f4308759e02feffbb8645d6983d9093292
汇总 SHA-256: ab607e749da1eedcf88441c5b6fd0db55e2ebae0bbd91d6d18e508605eef625b
协议 SHA-256: 5d53ccbd5a00763908162f0b700c0a3236e0d4f9d43b08e42ee31a2ef01aefbf

下载

结果 · CSV ↓协议 · JSON ↓核心矩阵 · JSON ↓

其他协议

全部协议 → · 首页上的核心矩阵 → · 全部数据集 → · 全部方法 →