Please use this identifier to cite or link to this item:
|Title:||EEG-based emotion recognition using deep learning||Authors:||Samriddhi, Govil||Keywords:||Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
Engineering::Electrical and electronic engineering::Control and instrumentation::Medical electronics
|Issue Date:||2022||Publisher:||Nanyang Technological University||Source:||Samriddhi, G. (2022). EEG-based emotion recognition using deep learning. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/158193||Project:||A3269-211||Abstract:||Emotion recognition is critical in both human-machine interfaces and brain-computer interfaces. Emotion is one of the most important factors in understanding human behavior and cognition. By precisely analyzing human emotion from electroencephalograms (EEGs) through methods such as deep learning and other traditional methods, we can extend this research to fields such as neural technology, cognitive science and psychology research. Furthermore, this can be utilized to create devices or software for assistance for people suffering from mental and cognitive disorders. Special focus needs to be given to the subject independent domain in order to increase the practicality quotient of such technology. This has proven to be difficult due to the varied nature of brain signal patters from one person to another. Through this project we have analyzed the current methods available for data pre-processing, feature extraction and classification in the emotion recognition domain. Signal pre-processing through down sampling and discrete wavelet transform have been performed in this report. Shannon entropy and wavelet energy were chosen as features for feature extraction. Dimension reduction was implemented through the use principal component analysis. Finally, the data was classified using baseline models consisting of a convolutional neural network and a long short-term memory network. Novel approaches were designed consisting of an ensemble network and a meta stack model network. Special sanity check was conducted to ensure the test predictions are subject independent. The CNN model upon generalization provided the best testing accuracy of 71.11% and the Meta Model ranked second with 66.67%.||URI:||https://hdl.handle.net/10356/158193||Schools:||School of Electrical and Electronic Engineering||Research Centres:||Fraunhofer Singapore||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Updated on Sep 18, 2023
Updated on Sep 18, 2023
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.