Cho2017 or GigaDB

Ref: Cho et al. (2017)

1 Dataset Overview

Field	Details
Modality	EEG + EMG
Paradigm/task	Cued left/right-hand MI
Participants	52 healthy
Channels	64 EEG + 2 EMG
Sampling rate	512 Hz
Classes/events	2 classes
Sessions/runs	1 session, 5 or 6 runs
Trials	100 or 120 trials/class (200 or 240/participant)
Trial/epoch duration	3 s MI period
Hardware/montage	BioSemi ActiveTwo, CMS/DRL active electrodes, standard_1005 BCI2000
License	CC BY 4.0
Data/source	Data Paper
MOABB	Cho2017

Note: subjects 31, 45, and 48 are usually removed in the literature; the source paper’s own analysis used 8–30 Hz band-pass and 10-fold CV with 7/3 train/test splits.

1.1 Montage and Trial Paradigm

Source: Cho et al. (2017)

Source: Cho et al. (2017)

Random Chance Cho et al. (2017)

The literature showed that the upper confidence limits of chance with α = 5% were 70% (classification accuracy) in a 2-class prob- lem, with 10 trials for each class.

2 Results

2.1 Summary

2.1.1 Protocol-Unclear Benchmark Extract

The local extract from Chevallier et al. (2024) does not identify enough protocol detail to assign these results without guessing.

pipeline	Cho2017
ACM+TS+SVM	73.56±14.54
CSP+LDA	71.38±14.54
CSP+SVM	71.92±14.25
DLCSPauto+shLDA	71.16±14.53
DeepConvNet	71.67±12.91
EEGITNet	57.20±12.21
EEGNeX	53.28±10.60
EEGNet-8,2	66.79±16.34
EEGTCNet	58.34±12.63
FBCSP+SVM	67.91±15.63
FgMDM	72.90±12.70
LogVar+LDA	64.49±10.08
LogVar+SVM	65.46±11.71
MDM	63.39±13.69
ShallowConvNet	73.84±14.95
TRCSP+LDA	71.85±13.84
TS+EL	76.23±14.21
TS+LR	75.01±13.71
TS+SVM	74.62±14.19
Average	68.47

2.2 Within-Session / Within-Subject

2.2.1 Cho Et Al. (2017) — 10-Fold Evaluation

The mean accuracy of all BCI performances (Fig. 4C) over the 50 subjects, excluding bad subjects (29 and 34), was 67.46% (±13.17%) in our datasets. Using Common Spatial Pattern (CSP) to extract 2 spatial filters as feature extraction method, and Fisher’s Linear Discriminant Analysis (FLDA) as classifier.
Source: Cho et al. (2017)

2.2.2 Kumar Et Al. (2019) — 10×10-Fold Cross-Validation

The source describes these results as subject-dependent 10×10-fold cross-validation misclassification rates. Cho2017 contains one recording session.

Subject	CSP	DFBCSP	SBLFB	SFTOFSRC	OPTICAL*	OPTICAL
1	27.30 ± 12.91	37.20 ± 9.54	37.20 ± 9.85	40.17 ± 10.63	20.70 ± 8.33	20.00 ± 9.37
2	49.00 ± 9.31	50.20 ± 11.56	44.10 ± 12.65	58.33 ± 8.64	45.70 ± 11.47	47.67 ± 12.37
3	11.40 ± 6.70	33.00 ± 10.15	8.40 ± 7.25	11.50 ± 6.97	8.60 ± 6.15	6.00 ± 5.78
4	39.10 ± 9.41	49.40 ± 13.46	19.40 ± 8.84	20.83 ± 10.99	24.10 ± 8.67	22.00 ± 10.05
5	1.00 ± 2.02	1.30 ± 2.22	1.00 ± 2.02	0.50 ± 1.53	0.90 ± 1.94	1.00 ± 2.03
6	20.20 ± 9.31	20.20 ± 8.02	17.50 ± 8.16	20.67 ± 6.40	16.60 ± 7.66	17.67 ± 7.74
7	50.42 ± 7.68	54.67 ± 10.02	45.17 ± 8.96	48.33 ± 10.92	47.75 ± 11.33	47.22 ± 7.92
8	49.60 ± 11.20	55.00 ± 11.78	53.90 ± 11.97	58.00 ± 9.43	43.80 ± 11.45	44.33 ± 11.65
9	49.50 ± 9.77	49.42 ± 11.67	44.33 ± 9.92	44.44 ± 9.75	43.75 ± 10.95	43.60 ± 10.07
10	42.70 ± 11.62	58.30 ± 8.84	36.40 ± 10.05	48.00 ± 9.06	30.90 ± 11.77	27.00 ± 10.31
11	45.80 ± 9.55	41.70 ± 9.98	49.80 ± 10.15	45.33 ± 7.98	49.30 ± 11.20	42.67 ± 9.89
12	31.00 ± 9.85	42.30 ± 8.82	37.40 ± 10.26	45.00 ± 10.51	34.50 ± 10.61	32.83 ± 8.97
13	11.10 ± 5.92	46.30 ± 9.57	11.50 ± 8.03	18.33 ± 8.64	14.60 ± 7.75	15.50 ± 8.24
14	4.80 ± 4.28	35.60 ± 10.63	5.20 ± 5.34	6.17 ± 5.20	4.70 ± 4.56	3.50 ± 3.51
15	49.90 ± 10.47	51.80 ± 11.94	33.20 ± 11.77	48.83 ± 13.75	43.00 ± 14.29	35.33 ± 11.29
16	51.50 ± 11.92	48.30 ± 11.00	51.60 ± 10.02	51.33 ± 10.25	52.10 ± 10.26	52.50 ± 11.89
17	51.80 ± 8.68	48.80 ± 9.88	47.20 ± 10.79	47.50 ± 10.06	50.20 ± 12.78	50.33 ± 9.46
18	49.00 ± 11.61	51.90 ± 12.81	41.10 ± 11.31	50.83 ± 9.83	49.30 ± 8.75	46.67 ± 8.34
19	45.50 ± 11.35	40.20 ± 9.20	35.40 ± 10.64	47.00 ± 12.36	42.90 ± 10.79	42.83 ± 9.16
20	36.20 ± 9.61	42.70 ± 11.83	48.70 ± 10.73	47.17 ± 11.12	28.30 ± 10.67	27.17 ± 10.14
21	45.70 ± 11.25	38.30 ± 10.53	35.20 ± 10.83	40.33 ± 9.73	40.20 ± 10.15	34.00 ± 11.33
22	45.70 ± 10.93	45.70 ± 9.69	43.40 ± 11.71	44.00 ± 11.33	46.60 ± 9.71	41.33 ± 10.08
23	32.60 ± 9.75	32.50 ± 10.66	25.40 ± 7.06	24.17 ± 9.11	19.80 ± 9.15	15.83 ± 8.31
24	53.30 ± 11.41	54.50 ± 12.71	49.40 ± 9.18	50.50 ± 9.94	46.90 ± 10.97	39.33 ± 10.23
25	56.70 ± 10.72	53.00 ± 12.78	46.90 ± 11.47	47.67 ± 10.73	49.00 ± 9.74	47.00 ± 14.18
26	4.20 ± 4.21	4.30 ± 3.91	3.30 ± 3.73	5.00 ± 3.94	3.70 ± 4.02	3.17 ± 3.34
27	52.50 ± 11.21	44.90 ± 11.72	51.00 ± 10.50	46.50 ± 11.00	57.00 ± 10.40	55.33 ± 9.99
28	20.30 ± 8.60	24.80 ± 7.49	21.00 ± 8.08	24.17 ± 10.35	19.50 ± 8.28	19.17 ± 4.93
29	54.80 ± 9.95	53.00 ± 11.87	57.70 ± 10.46	52.00 ± 10.55	56.30 ± 11.15	57.00 ± 10.88
30	44.00 ± 9.53	47.80 ± 10.16	40.70 ± 11.95	45.67 ± 10.06	44.80 ± 9.42	44.50 ± 10.45
31	45.00 ± 10.35	52.90 ± 11.78	38.20 ± 12.32	38.33 ± 6.61	38.60 ± 10.45	37.67 ± 10.97
32	49.60 ± 11.01	52.00 ± 10.93	51.60 ± 12.27	49.17 ± 10.67	49.60 ± 13.47	49.43 ± 11.99
33	48.90 ± 10.80	45.90 ± 10.63	46.20 ± 10.18	51.67 ± 11.24	47.20 ± 10.60	44.33 ± 12.23
34	44.10 ± 12.02	46.40 ± 9.95	45.60 ± 10.03	45.83 ± 10.35	42.70 ± 9.96	42.00 ± 7.50
35	18.90 ± 6.87	23.30 ± 8.96	27.70 ± 11.66	25.33 ± 9.91	18.40 ± 8.72	18.17 ± 9.33
36	47.30 ± 12.50	46.70 ± 9.72	44.90 ± 11.85	48.67 ± 10.82	33.70 ± 12.49	30.50 ± 10.20
37	26.60 ± 8.30	26.90 ± 9.94	26.00 ± 8.81	29.00 ± 8.65	23.80 ± 8.12	23.00 ± 10.20
38	53.40 ± 10.66	51.10 ± 10.02	51.70 ± 9.35	53.00 ± 9.52	52.40 ± 12.09	51.50 ± 12.12
39	28.60 ± 9.26	41.50 ± 10.22	29.80 ± 11.82	37.00 ± 9.25	28.20 ± 11.33	27.00 ± 9.25
40	48.60 ± 9.32	47.50 ± 10.99	53.10 ± 10.25	47.17 ± 7.62	48.40 ± 10.81	48.83 ± 9.16
41	25.70 ± 9.04	10.40 ± 6.05	19.60 ± 10.39	27.50 ± 7.74	15.90 ± 7.47	14.67 ± 8.80
42	56.00 ± 7.95	57.00 ± 10.83	51.00 ± 10.97	54.50 ± 8.84	51.10 ± 10.51	51.67 ± 10.11
43	6.70 ± 5.31	3.80 ± 4.47	3.60 ± 4.52	2.00 ± 2.82	4.90 ± 4.79	4.17 ± 4.17
44	10.60 ± 6.11	9.80 ± 6.54	10.70 ± 6.47	17.00 ± 8.37	8.40 ± 7.45	10.83 ± 7.55
45	46.20 ± 9.18	49.20 ± 10.27	49.40 ± 12.60	55.50 ± 11.62	45.70 ± 8.45	47.50 ± 10.97
46	30.58 ± 7.46	35.58 ± 9.72	24.92 ± 9.16	31.81 ± 9.44	24.67 ± 8.07	25.42 ± 8.71
47	27.10 ± 11.21	25.10 ± 8.36	29.10 ± 8.79	30.83 ± 11.07	22.00 ± 8.02	25.83 ± 9.83
48	44.00 ± 10.69	11.80 ± 7.68	19.00 ± 8.45	30.00 ± 8.41	35.30 ± 11.31	21.83 ± 9.69
49	10.50 ± 7.16	13.00 ± 6.70	10.40 ± 4.93	12.50 ± 8.59	13.70 ± 6.91	12.50 ± 6.66
50	0.00 ± 0.00	0.00 ± 0.00	0.00 ± 0.00	0.00 ± 0.00	0.00 ± 0.00	0.00 ± 0.00
51	50.40 ± 9.52	43.10 ± 11.29	47.10 ± 9.69	45.67 ± 9.89	47.90 ± 11.39	46.83 ± 10.71
52	42.50 ± 9.75	49.60 ± 10.59	39.50 ± 9.81	43.00 ± 9.15	40.10 ± 10.86	38.17 ± 13.10
Average	36.31 ± 9.14	38.46 ± 9.62	33.88 ± 9.38	36.80 ± 9.06	33.23 ± 9.38	31.81 ± 9.06

Source: Kumar et al. (2019)

2.3 Online / Real-Time

2.3.1 Kumar Et Al. (2019) — Real-Time Simulation

The source reports a simulated three-class real-time implementation. Values are misclassification rates (%).

Subject	Real-time
1	15.00 ± 9.05
2	23.25 ± 4.26
3	2.75 ± 2.19
4	14.50 ± 5.50
5	0.75 ± 1.21
6	8.00 ± 6.32
7	31.67 ± 7.20
8	24.00 ± 4.28
9	27.50 ± 4.89
10	14.25 ± 6.02
11	26.00 ± 6.69
12	15.75 ± 7.55
13	9.00 ± 2.93
14	1.50 ± 2.42
15	18.25 ± 7.91
16	26.75 ± 6.67
17	24.00 ± 4.12
18	24.00 ± 5.55
19	22.25 ± 5.95
20	12.00 ± 4.97
21	16.75 ± 6.46
22	20.75 ± 3.55
23	9.75 ± 3.22
24	23.25 ± 5.14
25	23.00 ± 4.97
26	1.50 ± 1.75
27	28.50 ± 5.30
28	12.00 ± 2.58
29	31.25 ± 7.38
30	22.25 ± 4.63
31	21.00 ± 3.57
32	24.75 ± 4.63
33	24.25 ± 5.66
34	24.50 ± 3.50
35	10.00 ± 2.89
36	19.50 ± 6.21
37	13.50 ± 6.15
38	33.50 ± 3.76
39	20.00 ± 6.12
40	31.00 ± 6.69
41	8.00 ± 2.30
42	27.25 ± 6.06
43	2.50 ± 2.36
44	5.25 ± 3.22
45	31.25 ± 8.76
46	12.70 ± 5.05
47	12.75 ± 5.46
48	11.50 ± 5.80
49	15.00 ± 6.12
50	0.00 ± 0.00
51	27.00 ± 5.11
52	19.00 ± 4.89
Average	17.78 ± 4.90

Source: Kumar et al. (2019)

2.4 Cross-Subject

2.4.1 Rodrigues Et Al. (2019) — RPA Transfer Learning

Results from: Rodrigues et al. (2019)

Dataset	N	MEAN AUC - DCT	MEAN AUC - RCT	MEAN AUC - PRL	MEAN AUC - OPT	MEAN AUC - RPA
Cho2017	1	0.54	0.59	0.58	0.57	0.54
Cho2017	5	0.55	0.61	0.61	0.57	0.59
Cho2017	10	0.55	0.62	0.62	0.57	0.62
Cho2017	25	0.57	0.64	0.64	0.58	0.66

2.4.2 Ouahidi Et Al. (2024) — Cross-Subject Ablation Study

The paper explicitly states that this ablation is conducted under the cross-subject evaluation paradigm. The Online row is its simulated-online ablation variant.

Results from: Ouahidi et al. (2024)

Pipeline	BNCI	Zhou	Physionet	Cho	Average Gain
Full pipeline	72.1 ± 7.3	81.8 ± 1.2	64.6 ± 4.4	75.4 ± 4.6	—
- BN	67.1 ± 8.4	77.3 ± 2.9	62.8 ± 4.2	74.5 ± 5.5	−3.0
- EA	66.7 ± 6.5	80.5 ± 1.3	63.0 ± 4.6	74.6 ± 4.6	−2.3
- Session	70.4 ± 7.2	79.9 ± 1.4	—	—	−1.8
Online	58.4 ± 6.9	73.1 ± 6.6	61.0 ± 4.9	73.5 ± 5.3	−6.8
- Mixup	69.1 ± 9.2	80.1 ± 0.7	60.0 ± 4.2	73.8 ± 5.2	−2.7
- Reg S	71.4 ± 7.9	81.8 ± 1.0	63.6 ± 4.1	74.2 ± 4.4	−0.72
- Everything	56.4 ± 9.0	74.5 ± 3.6	57.4 ± 4.6	72.6 ± 4.9	−8.3
+ EOG	79.8 ± 5.8	81.9 ± 1.6	—	—	+3.9
- LMSO	—	—	65.0 ± 14.3	76.2 ± 10.3	+0.6