Adaptive Parameter based Mantis Search Algorithm and Regularized Dropout with Gated Recurrent Unit for Multimodal Biometric Recognition
International journal of intelligent engineering and systems · 2025
Recently, biometric technologies are widely used for access control due to their uniqueness and their growing global demand has led to the development of biometric systems that integrate numerous features.However, unimodal biometric recognition faces challenges such as poor recognition accuracy due to inadequate discriminatory data.To overcome these limitations, multimodal biometric recognition combines face, fingerprint and iris modalities to enhance robustness and accuracy.This research proposes an Adaptive Parameter based Mantis Search Algorithm (APMSA) and Regularized Dropout with Gated Recurrent Unit (RDGRU) for multimodal biometric recognition.The APMSA is used to select significant features that enhance convergence speed, avoid local optima and increase global exploration.The RDGRU effectively models temporal dependencies, enhancing its ability to capture sequential changes in multimodal biometric recognition.The APMSA-RDGRU achieves an accuracy of 99.83% on the SDUMLA-HMT dataset, outperforming the existing Teacher Learning based Deep Neural Networks (TL-DNN).