Epilepsy cannot be underestimated as it can negatively impact every one of all ages and reduce the quality of life. Epilepsy can lead to sudden tumble and loss of awareness or consciousness, disturbances of movements. Fortunately, epilepsy seizures can be controlled if epilepsy is detected and treated properly. One of the widely used methods for detecting and diagnosing epilepsy is monitoring and analyzing electroencephalogram (EEG) signals. However, the traditional methods of monitoring and analyzing EEG have some challenges such as high costs, requirements of experienced medical experts, non-scalability, or non-support real-time and long-term monitoring. Therefore, in this paper, we present an advanced deep learning neural network approach for the automatic detection of epilepsy seizures. The proposed approach with a customized attention mechanism can be used for a single EEG channel. We evaluate the approach with the Bonn dataset and the CHB-MIT dataset and achieved higher than 98% accuracy, 99% sensitivity, and 98% specificity for a single EEG channel in most of the cases. The results show that the proposed approach is a potential candidate for enhancing automatic epileptic seizure detection systems.