Abstract: Recently, biometric authentication systems have become essential tools for secure and reliable identity verification. Electrocardiogram (ECG) biometrics are particularly effective because of ...

Abstract: Feature extraction and selection in the presence of nonlinear dependencies among the data is a fundamental challenge in unsupervised learning. We propose using a Gram-Schmidt (GS) type ...

The proliferation of open-sourced Large Language Models (LLMs) and diverse downstream tasks necessitates efficient model selection, given the impracticality of fine-tuning all candidates due to ...