Electroencephalography (EEG) is a complex signal and can require several years of training to be correctly interpreted. Recently, deep learning (DL) has shown great promise in helping make sense of EEG signals due to its capacity to learn good feature representations from raw data. Whether DL truly presents advantages as compared to more traditional EEG processing approaches, however, remains an open question. Emotion- or affect-recognition is the task of estimating emotional responses
to some stimulus. Humans typically recognize emotion by integrating multiple
modalities, whether from vocal intonation, perspiration, or facial expressions.
Different modalities may also offer different benefits and challenges than one
another for affect recognition, both in terms of their descriptive power, but also
in terms of how practical or economical they are to capture.
Electroencephalographs (EEGs) measure activity in the brain by amplifying
electrical brain waves created in psycho-physiological processes such as experi-
encing an emotion. One motivation for using EEG for affect recognition is its use
for augmented human-computer interfaces, especially in instances where physical
impairments prevent the use of traditional interfaces.
In this work we present a comparision of different approaches to use DL in EEG processing. Deep learning network(DLN) is capable of discovering unknown feature coherencesof input signals that is crucial for the learning task to representsuch a complicated model. The DLN provides hierarchicalfeature learning approach. Learned features at high-level arederived from features at low-level with greedy layer-wiseunsupervised pre-training. This unsupervised pre-trainingprovides the stage for a final training phase that is fine-tuning

Reading the articles showed us several approaches to neural networks. This allowed us to understand their structure. On the basis of presented articles, we've chosen Human Emotion Recognition with Electroencephalographic Multidimensional Features by Hybrid Deep Neural Networks as a most interesting one. Nevertheless, all occurring ‘networks’ in fact have generally build in common. Main differences are in the model of each layer. Networks are using combinations of models like hybrid deep learning model, recurrent neural network, which have big impact on their performance. We can also observe more differences like using different sampling on the input for each model. It is impossible to choose a state of art based on attached articles because some of them like “Deep Physiological Affect Network for the Recognition of Human Emotions” are comparing to general “state-of-art” In fact they are kinda outdated. As we mention in the first words of this summary we found the most interesting model of network basing on its results.. The next thing in common is the joint model of the dataset used in experiments called DEAP using it made results more comparable.