Dual Kernel Support Vector-based Crossover Red Fox Algorithm: Advancements in Assistive Technology for Hearing-impaired Individuals

Abstract

Individuals with hearing impairment face several challenges, including difficulties in communication, social interactions, and accessibility to information on various auditory abilities. Innovations range from hearing aids to advanced communication devices and mobile apps. Designing solutions that prioritize user feedback ensures a more inclusive and empowering experience for people with hearing impairment. Assistive technology (AT) endeavors to enhance the daily lives of individuals, fostering greater connectivity, and also plays a crucial role in addressing these challenges. Therefore, an attention dual kernel support vector-based crossover red fox (ADKS-CRF) algorithm is developed for superior performance. This research study proposes a model combining a dual kernel support vector machine with an attention mechanism to implicitly operate in a high-dimensional feature space without computing the transformed vector features. The crossover strategy is incorporated with the red fox optimization algorithm, and the integrated formation of CRF fine-tunes the parameters of the ADKS model, removing the complexity of local optima. This work conducted experiments using raw data from an effective 3D ear acquisition system dataset. Experimental validation is conducted using various evaluation measures to assess effectiveness. The proposed hybrid approach achieves a sensitivity of 97.8%, an F1-score of 97.1%, a specificity of 96.3%, an accuracy of 98.4%, a false alarm rate of 90.8%, and a Matthews correlation coefficient of 97.3%. The comparative analysis evaluates the efficacy of the ADKS-CRF method with various baseline approaches for the development of ATs for hearing-impaired people.