Table 7.

Binary sentiment analysis results of the multi-modal model using the convolutional EEG decoding component.

	Randomly initialized			GloVe			BERT
Model	P	R	F1 (std)	P	R	F1 (std)	P	R	F1 (std)
Baseline	0.572	0.573	0.552 (0.07)	0.751	0.738	0.728 (0.08)	0.900	0.899	0.893 (0.04)
+ noise	0.558	0.584	0.528 (0.11)	0.780	0.767	0.762 (0.06)	0.895	0.887	0.883 (0.05)
+ ET	0.617	0.623	0.610 (0.07)	0.790	0.790	0.783 (0.06)	0.896	0.887	0.881 (0.05)
+ EEG full	0.588	0.583	0.572 (0.04)	0.778	0.774	0.772+ (0.05)	0.928	0.927	0.926* 0.03
+ EEG θ	0.564	0.569	0.535 (0.08)	0.805	0.792	0.791+ (0.04)	0.922	0.919	0.917* (0.03)
+ EEG α	0.596	0.593	0.563 (0.08)	0.775	0.781	0.772* (0.08)	0.920	0.917	0.916* (0.03)
+ EEG β	0.605	0.597	0.580 (0.08)	0.802	0.797	0.792+ (0.05)	0.920	0.914	0.914* (0.04)
+ EEG γ	0.640	0.625	0.611+ (0.09)	0.787	0.780	0.776+ (0.05)	0.905	0.905	0.901 (0.04)
+θ+α+β+γ	0.599	0.579	0.558 (0.07)	0.800	0.794	0.786+ (0.05)	0.909	0.910	0.907 (0.04)

We report precision (P), recall (R), F₁-score and the standard deviation (std) between five runs. The best results per column are marked in bold. Significance is indicated on the F₁-score with asterisks:

^*denotes p < 0.05 (uncorrected), + denotes p < 0.003 (Bonferroni corrected p-value).