Fig. 5From: CNN-based framework using spatial dropping for enhanced interpretation of neural activity in motor imagery classificationRelevance weights computed for representative individuals (best-performing A03T and A08T and worst-performing A02T and A05T) in terms of performed three-class accuracy for scenario C: without spatial dropping (left column), spatial dropping by weighing only all sensorimotor electrodes (middle column), spatial dropping by excluding nonrelevant electrodes (right column)Back to article page