Series:

SpringerBriefs in Computational Intelligence

2 total works

Cover of AI, Ethical Issues and Explainability—Applied Biometrics
AI has contributed a lot and biometrics is no exception. To make AI solutions commercialized/fully functional, one requires trustworthy and explainable AI (XAI) solutions while respecting ethical issues. Within the scope of biometrics, the book aims at both revisiting ethical AI principles by taking into account state-of-the-art AI-guided tools and their responsibilities i.e., responsible AI. With this, the long-term goal is to connect with how we can enhance research communities that effectively integrate computational expertise (with both explainability and ethical issues). It helps combat complex and elusive global security challenges that address our national concern in understanding and disrupting the illicit economy.

Cover of Active Learning to Minimize the Possible Risk of Future Epidemics

Future epidemics are inevitable, and it takes months and even years to collect fully annotated data. The sheer magnitude of data required for machine learning algorithms, spanning both shallow and deep structures, raises a fundamental question: how big data is big enough to effectively tackle future epidemics? In this context, active learning, often referred to as human or expert-in-the-loop learning, becomes imperative, enabling machines to commence learning from day one with minimal labeled data. In unsupervised learning, the focus shifts toward constructing advanced machine learning models like deep structured networks that autonomously learn over time, with human or expert intervention only when errors occur and for limited data—a process we term mentoring. In the context of Covid-19, this book explores the use of deep features to classify data into two clusters (0/1: Covid-19/non-Covid-19) across three distinct datasets: cough sound, Computed Tomography (CT) scan, and chest x-ray (CXR). Not to be confused, our primary objective is to provide a strong assertion on how active learning could potentially be used to predict disease from any upcoming epidemics. Upon request (education/training purpose), GitHub source codes are provided.