The detection of epileptic seizures by electroencephalography (EEG) signals has become a standard method in recent years for the diagnosis of epilepsy. Accurate and automatic detection of epileptic seizures is needed since manual identification of epileptic seizures by specialist neurologists is a time-consuming and labor-intensive process, which also leads to various errors. For this purpose, frequency-based features were extracted from the EEG signal and various classifiers based on ensemble learning were used to detect epileptic seizures automatically. The performance of the proposed method was tested using cross-validation and cross-patient experiments. According to the experimental results, sensitivity, specificity, and accuracy rates were obtained 94%, 93% and 93% for cross-validation and 76%, 90% and 90% for cross-patients, respectively.