Deep SSVEP Convolutional Unit (SCU)

Fig 1: Block diagram of the DeepSCU classification algorithms. The green box and red box represent a single SCU and dense block respectively.^[1]

SCU block

Each SCU block is composed of the following layers:

• 1D Convolutional layer
  To be performed on the EEG samples. Returning a feature map based on the time window and number of filters chosen.
• Batch Normalization layer
  To keep the average and the standard deviation of the output close to 0 and 1, respectively.
• Max Pooling layer
  To take the maximum value on a spatial window of size equal to 2.
• Rectifier Linear Unit (ReLU) activation function
  To return 0 if the input is negative, otherwise return the reveived value of input as output.

References