Fig. 6
From: A multi-task EEG emotion recognition method based on emotion-dimension coupling constraints
Performance comparison of ablation studies on (A) the DEAP dataset and (B) the DREAMER dataset. STL denotes the single-task model, while MTL represents the multi-task framework. “+None” indicates the absence of constraints; “+1”, “+2”, and “+3” correspond to the application of the V–A circular geometric constraint, the A–D energy alignment constraint, and the V–D correlation constraint, respectively. Combined notations such as “MTL+1+2” denote the simultaneous use of multiple constraints, and “MTL+All” applies all three constraints.
