???????????????????????????????????????????? Consequently, it is the greatest exercise to freeze all layers in the ParallelConv1D blocks and only high-quality-tune the LSTM levels as well as the classifier without unfreezing the frozen levels (situation two-a, along with the metrics are proven in

Article Under Review