Machine Learning For Classification Of Antithetical Emotional States

TL;DR

This paper evaluates traditional machine learning classifiers and a tabular deep learning approach for emotion classification using EEG signals, addressing data scarcity and feature learning challenges to improve accuracy.

Contribution

It introduces a tabular learning approach that achieves state-of-the-art results without heavy neural networks, enhancing EEG-based emotion classification.

Findings

Tabular learning outperforms traditional classifiers on DEAP dataset.

Deep learning architecture boosts performance without heavy neural networks.

Addressed data scarcity and feature learning issues in emotion classification.

Abstract

Emotion Classification through EEG signals has achieved many advancements. However, the problems like lack of data and learning the important features and patterns have always been areas with scope for improvement both computationally and in prediction accuracy. This works analyses the baseline machine learning classifiers' performance on DEAP Dataset along with a tabular learning approach that provided state-of-the-art comparable results leveraging the performance boost due to its deep learning architecture without deploying heavy neural networks.