Dataset: Epilepsy data has been taken from University of Bornn, germany. Dataset consist of 1D signals time series data. The complete data consist of five sets (A to E), each containing 100 single channel instances. Set A consist of EEG signals recorded from healthy volunteers while they were in relaxed and awake state with eyes opened. Whereas, the EEG signals in set E were recorded only during seizure activity.
We were able to achieve the accuracy of 99.09%.
Input_1 (InputLayer) (None, 512, 1)
Conv1d_1 (Conv1D) (None, 24, 5, 3)
batch_normalization_1 ()
Activation('relu')
Conv1d_2 (Conv1D) (None, 16, 3, 2)
batch_normalization_2 ()
Activation('relu')
Conv1d_3 (Conv1D) (None, 8, 3, 2)
batch_normalization_3 ()
Activation('relu')
flatten_1 (Flatten)
dense_1 (Dense) (None, 20, activation="relu", kernel_initializer = "uniform")
dense_2 (Dense) (None, 1, activation = 'sigmoid')