Efficient HyperFused Model for Stratification of Concurrent Facial Expressions

Abstract

This article represents a novel method called HyperFusion that amalgamates Principal Component Analysis (PCA) with Support Vector Machines (SVM) to enhance classification precision and computational competence in facial expression recognition which involves high-dimensional datasets. HyperFusion leverages PCA for dimensionality reduction by extracting the most significant features in stratifying the concurrent facial expressions, while discarding irrelevant or redundant data. This reduction streamlines the feature space, making the SVM classifier more effective by mitigating the curse of dimensionality and improving generalization. SVM then classifies the data using an optimal hyperplane in the lower-dimensional subspace generated by PCA. The fusion of these two methods results in faster training times, reduced complexity, and improved classification accuracy. Experimental results demonstrate that HyperFusion achieves an outstanding performance concerning accuracy and computational efficiency compared to traditional SVM classifiers applied to untransformed high-dimensional data. The proposed method HyperFusion produces 90.3 % accuracy for CK+ dataset. This approach is particularly well-suited for tasks involving large feature sets, noisy data, and applications where real-time processing is critical.

Topics & Concepts