Fig. 1From: CNN-based framework using spatial dropping for enhanced interpretation of neural activity in motor imagery classificationThe proposed CNN structure that is based on Wide&Deep neural network handling multiple inputs. The first layer (IN1) is the input, the second (CN2) and third layers (MP3) are hidden and accomplish the feature mapping generation, while the next block (ranging from the output of layer CT4 to the OU9 layer) comprises the classification stageBack to article page