A newly developed Adaptive Fuzzy–Cost Benefit Analysis (AF–CBA) framework has demonstrated significant improvements in classification performance and reduced misclassification costs for electroencephalography (EEG) cognitive state classification compared with baseline approaches, with an optimized J48 decision tree classifier achieving an accuracy of 99.95%.
Conventional EEG-based classification techniques rely on fixed-threshold and accuracy-driven optimization, which produce unstable decision boundaries and higher misclassification costs under cognitively ambiguous conditions, alongside narrow performance differences across standard classifiers. To address these constraints, the proposed AF–CBA framework incorporates fuzzy inference to model emotional resilience and thinking ability as structured cognitive indicators, generating transparent cognitive load features that are subsequently evaluated across machine learning algorithms including Decision Tree (J48), Random Forest (RF), Support Vector Machine (SVM), and Naive Bayes.
In baseline experimental evaluations, the tested classifiers demonstrated classification accuracies spanning from 80.21% to 96.85%, exhibiting a small performance gap among SVM, RF, and J48. To improve decision stability and mitigate false classifications, the AF–CBA framework applies cost-sensitive optimization, stochastic correction-based stabilization, and adaptive threshold refinement, yielding the peak 99.95% accuracy observed with the optimized J48 model.
The researchers report that the resulting framework provides an interpretable, adaptive, and cost-effective approach for EEG-based cognitive load classification. The framework has high potential for intelligent cognitive monitoring, adaptive neurocognitive systems, and real-time human-centred artificial intelligence applications.
Subject of Research: EEG cognitive state classification using an adaptive fuzzy cost-benefit framework
Article Title: A novel AF-CBA framework for accurate and interpretable EEG cognitive state classification
Article References: Mandal, B., Mohapatra, S., & Sahoo, R. (2026). A novel AF-CBA framework for accurate and interpretable EEG cognitive state classification. Neural Computing and Applications, 38(20), Article 840. https://doi.org/10.1007/s00521-026-12602-1
Image Credits: AI Generated
DOI: 10.1007/s00521-026-12602-1
Keywords: EEG classification, cognitive states, fuzzy inference, machine learning, decision tree, cost-benefit analysis, adaptive threshold optimization, cognitive load, novel, AF-CBA, framework, accurate
News Source: Denise Maddox. (November 7, 2026). Adaptive Fuzzy Framework Improves EEG Cognitive State Classification Performance and Interpretability. Scienmag.
