Consumer-friendly EEG-based Emotion Recognition System: A Multi-scale Convolutional Neural Network Approach

Tri Duc Ly, Gia H. Ngo · 2025

Read Consumer-friendly EEG-based Emotion Recognition System: A Multi-scale Convolutional Neural Network Approach by Tri Duc Ly, Gia H. Ngo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-16448.txt.

Introduction

A home-grade brain wave emotion reader is built by Tri Duc Ly and Gia Ngo in a 2025 arXiv paper. Dry sensors and cheap headsets now put EEG in homes. Their net reads feeling scores on three axes: pleasantness, arousal, and control feel. Kernels at many scales pull traits from the wave stream. One new kernel learns from four brain zones at once.

The model beat a known past best, TSception, across score gauges. Runs stayed steady across repeats of the tests. Home emotion tools gain a stronger core from the design.

Catalog record

CREATORTri Duc Ly, Gia H. Ngo
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-16448

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Consumer-friendly EEG-based Emotion Recognition System: A Multi-scale Convolutional Neural Network Approach Tri Duc Ly, Gia H. Ngo Abstract EEG is a non-invasive, safe, and low-risk method to record electrophysiological signals inside the brain. Especially with recent technology developments like dry electrodes, consumer-grade EEG devices, and rapid advances in machine learning, EEG is commonly used as a resource for automatic emotion recognition. With the aim to develop a deep learning model that can perform EEG-based emotion recognition in a real-life context, we propose a novel approach to utilize multi-scale convolutional neural networks to accomplish such tasks. By implementing feature extraction kernels with many ratio coefficients as well as a new type of kernel that learns key information from four separate areas of the brain, our model consistently outperforms the state-of-the-art TSception model in predicting valence, arousal, and dominance scores across many performance evaluation metrics. 1 Introduction Emotion is a fundamental part of human life. Associated with feelings, emotions are important to one’s decision-making, learning process, and many other cognitive processes [ 1 ] . They are also a key tool for human interaction and communication, serving as a way for the communicator to express themself, providing information on their state of mind, feelings, motives, and intentions [ 2 ] . Therefore, the interest in researching, learning, and further understanding human emotion and its impact has been growing, especially in the field of neuroscience [ 3 ] . While much progress is still to be made with research regarding the matter due to the complexity behind human emotion [ 4 ] , many advancements have been achieved thanks to the development of technology in the field, contributing to novel approaches that researchers can use to further study human emotion. With more knowledge of emotion, a key to perceiving feelings, expression, and cognitive information of a person, there are a number of ways this can be used to improve the quality of life, one of which, is therapy. In recent years, a number of technological advancements have led to an increased interest in a resource that can be used for the task: electroencephalography (EEG), one of the most widely used brain imaging technologies. EEG presents a non-invasive way to measure brain electrical activities, which can then be passed through a brain-computer interface (BCI) to further process the information and identify human emotion. This surge in interest is due to the development of consumer-grade EEG devices with dry electrodes. Before this development, despite its potential, the applications of EEG-BCI systems outside of research labs are extremely sparse due to the limitations of research-grade EEG devices: (1) it is time-consuming to set up a research-grade EEG device, typically taking from 30 to 60 minutes, (2) user’s mobility is heavily restricted due to the high number of wires, and (3) the extremely high cost of the devices [ 5 ] . Even though data recorded by research-grade EEG devices may provide more information and allow EEG-BCI systems to yield better results in the task of emotion recognition, consumer-grade EEG devices open up countless more research and consumer applications with their affordability, portability and simplicity, while still providing reliable results [ 6 ] . With automatic emotion recognition using EEG signals, the use of machine learning (ML) algorithms, specifically deep learning (DL), is one of the most popular and reliable methods due to its known ability to learn non-linear patterns from complex information - an aspect that is found in EEG. Thus, we developed a DL model inspired by TSception [ 7 ] that predicts human emotion based on EEG with reliable accuracy. Our method provides a novel approach to which a DL architecture can be used for the task. The remainder of this paper is organized as follows. In Section 2 , we give background information on emotion, EEG, EEG-based emotion recognition, and deep learning. Section 3 provides the details of our materials and method by introducing the dataset and the performance evaluation metrics that we used, the experimental setup of the DL model, and the mechanism behind our model. In Section 4 , we present the results of our model and analyze how these results can be interpreted. Future implications, suggestions, and a discussion regarding our study are given in Section 5 . Finally, the conclusion and guidance acknowledgment are presented in Sections 6 and 7 . 2 Background 2.1 Emotion Emotion has long been a complicated subject of research. One of the earliest attempts to explore the mechanism behind human emotions comes from the James-Lange theory of emotion in 1884 [ 8 ] , in which psychologist William James and psychologist Carl Lange suggested that emotion is the result of a physiological response to external stimuli (e.g., your body trembling and your heart beating rapidly would cause you to feel fear). This theory was challenged in 1927 by Walter Cannon and Phillip Bard, who argued in the Cannon-Bard theory that emotion and physiological response occur simultaneously, not consequently [ 9 ] . The Freudian theory by Sigmund Freud, one of the greatest and most well-known theories in psychology, introduced the idea that much of human behavior, including human emotion, is greatly influenced by the unconscious mind. In the 1950s, the cognitive revolution led to the development of novel theories such as the Schachter-Singer Two-Factor Theory of Emotion, which suggests that emotion consists of two components: physical arousal and a cognitive interpretation [ 10 ] . In the 1960s and the 1970s, Richard Lazarus pioneered the advancement of the cognitive appraisal theory, stating emotion is the response to an individual’s evaluation of a situation. As the 21st century approaches, advances in brain imaging like functional magnetic resonance imaging led to many studies about the brain’s role in human emotions, the emergence of the field of affective neuroscience, and the popularization of emotional intelligence. These advancements, with pioneers like Jaak Panksepp and Daniel Goleman, have led to rapid progress in extensive research on the role of emotion and the mechanism behind it. Emotion plays many roles in human life: (1) Emotions serve as social signals in human exchanges, conveying one’s feelings, perceptions, and situational understanding to the interaction partner [ 11 ] ; (2) Emotion greatly influences cognitive function, which includes decision making, perception, learning, and maintaining health [ 1 ] ; (3) Emotion significantly affects the course of actions, execution, control, and the way one explains their actions [ 12 ] . It also aids our daily lives in various ways. Most evidently, emotion is known to provide memory benefits, though it enhances memory more for negative experiences than positive ones [ 13 ] . As a social signal, it serves as a tool for humans to connect with others, navigating social interactions so that we can resonate with and care for another person [ 14 ] . The positive and negative nature of emotions can also affect human health, as well as influencing and experiencing work processes [ 15 ] , [ 16 ] . There is no unitary definition for emotion [ 17 ] , and the way emotion is perceived has vastly varied throughout modern science. The way people feel certain emotions can be different from how others do, and the description

Read the complete text