Security risks and hacks have increased over the years. With authentication being an important part of security, electroencephalography or the measurement of brain signals has emerged as a new biometric form of authentication. In previous studies, it has been shown that EEG differs from person to person. EEG-based user authentication systems have the advantages of both password-based and biometric-based authentication systems but without their downsides. Many studies have been done in this field, however, a comparison between different classification methods for authentication of brain waves has not yet been proposed. This study aimed to find the classifier with the most accurate results. The classifiers used are artificial neural network, support vector machines, k-nearest neighbor, hidden Markov model, and the Gaussian mixture models. Moreover, this study will provide a clean set of EEG data of both imagined speech and images for public research. After comparing the performance of these algorithms using raw EEG data, the results show that the algorithms with the best performance are KNN and ANN with accuracy of 98.23% and 69.34% respectively. Followed by GMM with 56.81% accuracy and SVM with 54.50% accuracy. Lastly HMM with accuracy of 50.49%. The overall results show that raw EEG data is not discriminative enough to be used for authentication systems.