Optimizer Sensitivity In Vision Transformerbased Iris Recognition: Adamw Vs Sgd Vs Rmsprop

TL;DR

This paper investigates how optimizer choice affects the accuracy and stability of Vision Transformer models in iris recognition, highlighting the importance of optimizer selection for biometric system robustness.

Contribution

It provides a comparative analysis of AdamW, SGD, and RMSprop optimizers on ViT-based iris recognition, an area previously underexplored.

Findings

Optimizer choice significantly impacts recognition accuracy.

AdamW yields more stable training than SGD and RMSprop.

Insights help improve robustness of biometric identification models.

Abstract

The security of biometric authentication is increasingly critical as digital identity systems expand. Iris recognition offers high reliability due to its distinctive and stable texture patterns. Recent progress in deep learning, especially Vision Transformers ViT, has improved visual recognition performance. Yet, the effect of optimizer choice on ViT-based biometric systems remains understudied. This work evaluates how different optimizers influence the accuracy and stability of ViT for iris recognition, providing insights to enhance the robustness of biometric identification models.