Analyzing time series data of an electroencephalogram (EEG) is important for medicine and neuro- logical research. Due to noise and patient-specific signals it is difficult to analyze these data. Standard CNN, FNN and RNN in combination with noise filtering are commonly used in research. This review paper should give an overview over current variations of these neural networks applied on same datasets and evaluated on comparable experimental setups. A literature review was performed for this purpose, from which various methods are explained and results on same datasets presented. Overall all state-of-the-art methods perform better accuracy on difficulty tasks than basic RNN. The Temporal Convolutional Network (TCN) was able to achieve the highest average accuracy (82.97 ± 13.77 %) in all experiments. The temporal component of TCN holds promise for generic use in different brain wave measurements, for example to classify emotions.