The analysis of EEG signal have been widely used in the diagnosis process of neurological disease as epilepsy. Epilepsy, characterized by recurrent seizures, is one of the most common neurological disorder that affect people at all ages. It is an abnormal electrical activity in the brain. The electroencephalograph (EEG) can provide valuable insight and improved understanding of the mechanisms causing epileptic disorders. The detection of epileptiform discharges in the EEG is important to diagnose epilepsy. Before analyzing the EEG signal it was necessary to extract the important features to minimize the loss of useful information, but with the implantation of (DL) Deep Learning we can automatically from EEG raw data extract features and classify them. Deep learning is an artificial intelligence technique with powerful algorithms that deduplicates feature extraction, dimensionality reduction and classification steps, also DL methods have been demonstrated its effectiveness in accurately classifying time series data.This proposed paper focuses on three techniques of DL for automatic epileptic seizure identification which are (CCN) convolutional neural network model, (LSTM) long short-term memory model and (BiLSTM) bi-directional long short-term memory model in order to enhance the accuracy of epilepsy detection. To do we will use 5 categories of data and only the subjects in class 1 have epileptic seizure the other categories {2, 3, 4, 5} are healthy patients. Our motivation for classifying this version of data is to evaluate seizure activity versus normal activity, this Dataset collect from UCI Machine Learning Repository. The best performance of all implemented methods resulted in a BiLSTM model with accuracy of above 98.87%. This work was programmed in Python programming environment with Jupyter notebook.