A predictive method for emotional sentiment analysis by machine learning from electroencephalography of brainwave data
Küçük Resim Yok
Tarih
2022
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Elsevier
Erişim Hakkı
info:eu-repo/semantics/closedAccess
Özet
The human cerebrum is the focal handling unit for different assignments, for example, observation, comprehension, consideration, feeling, memory, and activity. Understanding of Brain-Computer Interface strategies and revamping human feelings is an exceptionally tremendous field in the area of exploration. Numerous studies were conceived to distinguish the human feelings as bliss; fear, outrage, and bitterness were discovered promising by utilization of electroencephalography (EEG) signs. Experimental datasets were collected from a Muse EEG headband with a global EEG position standard. This arrangement collects 2549 datasets based on time-frequency domain statistical features where a subset of 640 datasets chosen by their symmetrical uncertainty was discovered to be best when utilized with three different classifiers Random Forest (RF), XG Boost, and Decision Tree for emotion detection. All these three algorithms achieved an overall accuracy of more than 95%. XG Boost exhibits the maximum accuracy of 99.04% while RF takes the least training time and occupied the maximum Area under the curve. By considering the entire performance index it is seen that all the proposed algorithms outperform the state-of-art methods. © 2023 Elsevier Inc. All rights reserved.
Açıklama
Anahtar Kelimeler
Classification; Eeg Signal; Machine Learning; Performance Index; Symmetric Uncertainty
Kaynak
Implementation of Smart Healthcare Systems using AI, IoT, and Blockchain
WoS Q Değeri
Scopus Q Değeri
N/A
