Using an intelligent method to automatically detect epileptic seizures in medical applications is one of the most important challenges in recent years to reduce the workload of doctors in the analysis of epilepsy data through visual inspection. One of the problems of automatic detection of various epileptic seizures is the extraction of desirable characteristics, in such a way that these characteristics can make the most distinction between different phases of epilepsy. The process of finding the right features is usually a matter of time. This research presents a new approach for the automatic identification of epileptic episodes. In this paper, a deep convolutional network with eight convolutional layers and two fully-connected layers is provided to learn the characteristics hierarchically and automatically identify epileptic episodes using the EEG signal. The results show that the use of deep learning in applications such as learning characteristics hierarchically and identification of different stages of epilepsy has a higher success rate than other previous methods. The proposed model presented in this paper provides an average of 100% accuracy, sensitivity and specificity for the classification of three different epileptic seizures.