Revolutionizing EEG-based emotion recognition in achieving one health

EEG-based emotion recognition holds immense potential for applications in human-robot interaction, computer gaming, and mental health monitoring. However, existing methods face challenges such as high-dimensional data, computational complexity, and impractical experimental setups. To address these limitations, this research introduces a pyramidal structured feature extraction approach that leverages hierarchical forward interpolation and multirate sampling frequencies. A culturally relevant dataset was specifically designed to ensure ecological validity, incorporating Kannada musical clips and commercial advertisements as emotional stimuli. EEG signals were recorded from 46 participants using a BIOPAC 2-channel device, capturing emotional states such as enjoyment, relaxation, sadness, fear, and being scared. The proposed pyramidal approach effectively reduces dimensionality by condensing EEG features from thousands to a compact yet information-rich set. It employs iterative forward difference operations across five hi erarchical levels, while multirate sampling enhances temporal resolution for improved emotion discrimination. In the second stage, a probabilistic neural network (PNN) is employed, achieving classification accuracies of up to 98%, surpassing traditional principal and independent component analysis techniques.The implications of this work extend across multiple domains. In human-robot interaction, the proposed method enhances affective computing by enabling real-time adaptation of robotic responses based on users’ emotional states. In computer gaming, it facilitates the development of adaptive gaming experiences, where game difficulty, narratives, or virtual environments adjust dynamically to players’ emotions. For mental health monitoring, this approach offers a non-invasive tool for early detection and management of affective d isorders, e nabling personalized therapeutic interventions. Additionally, the integration of culturally relevant stimuli paves the way for region-specific e motion r ecognition models, e nsuring i nclusivity and improved a ccuracy in diverse populations. These contributions position the proposed pyramidal approach as a scalable and adaptable solution for EEG-based emotion recognition across real-world applications.

Citation

Thejaswini, M.S., Kumar, G. H., Manjunath Aradhya, V.N., Renukappa, S., Subbarao, C., Suresh, S., Veenith, T., Rudrapatna Sathyamurthy, S.C. (in press) Revolutionizing EEG-based emotion recognition in achieving one health. In Achieving Sustainability Using Artificial Intelligence and Digital Technologies: Perspectives from the Global South. Springer.

Publisher

Springer

URI

https://wlv.openrepository.com/handle/2436/626179

Type

Chapter in book

Language

en

Description

This is an author's accepted manuscript of a book chapter due to be published by Springer in Achieving Sustainability Using Artificial Intelligence and Digital Technologies: Perspectives from the Global South. For re-use please see Springer's terms and conditions.

Collections

Faculty of Science and Engineering