Deep Neural Based Learning of EEG Features Using Spatial, Temporal and Spectral Dimensions Across Different Cognitive Workload of Human Brain: Dimensions, Methodologies, Research Challenges and Future Scope

Training to Deep learning based model can be executed simultaneously in order to learn complicated information from several domains, in contrast to the traditional cognitive workload recognition paradigms. The earlier stated techniques typically collect characteristics from the spectral and temporal views of the data separately. Therefore, the crucial step in establishing reliable EEG representations related to the identification of cognitive workload is to select deep learning-based models. The prime objective of this study is to focus on those deep learning based techniques which are effective and efficient in the identification of cognitive workload using Electroencephalogram (EEG) signals. Human brain is a dynamic entity and holds the tendency of constantly thinking about the past and the future instead of relaxing in the present. Many models dealing with cognitive assessment fail to perform better and gave poor performance due to this property of human mind. As a result, it becomes essential and mandatory to remove tension and anxiety as a noise using a variety of techniques out of the collected EEG signals of the human brain. A thorough study based on implementing of different deep learning techniques in identification of EEG-based cognitive or mental workload was carried out to help the researchers and scientists in getting the complete information in a single platform. The entire study conducted briefly describes all the concepts of cognitive workload recognition using deep learning in a systematic manner. The study reveals that the CNN outperformed all the other deep learning techniques when used for conducting any kind of human brain cognitive state assessment whereas RNN gave poor performance.