In this present work, a unique strategy for discrimination of epileptic electroencephalogram (EEG) signals is contemplated utilizing autocorrelation technique. The EGG signals recorded from healthy as well as epileptic patients were initially split into five rhythms to examine their variations in time frame. Next, autocorrelation sequence of the individual rhythms was computed to investigate their self-similarity. Then from each EEG rhythm, four statistical features were extracted. The statistical features were used to categorize EEG signals employing support vector machines (SVM) classifier. It has been observed that using autocorrelation-based features, SVM can accurately discriminate healthy and epileptic brain rhythms with a high level of classification accuracy. Among different EEG rhythms, it was observed that the low frequency i.e., delta rhythm yielded 100% accuracy compared to high frequency sub-bands. Moreover, the efficacy of the proposed model was subsequently evaluated using two more machine learning classifiers such as k-nearest neighbor (kNN) and Naïve Bayes (NB). The box-plot visualization of the extracted features and the one-way analysis of variance (ANOVA) test were used to validate the significant distinguishability between two groups. The proposed framework can be implemented to develop a computer-aided detection system for detection of epilepsy.