Loading...
Arockia Raj A is an Assistant Professor in the Department of Computer Science and Engineering (Data Science) at Madanapalle Institute of Technology & Science, India. His research interests include Data Science, Machine Learning, Artificial Intelligence, and intelligent analytics.
The rapid expansion of digital financial services, internet banking, cloud computing, the Internet of Things (IoT), and e-commerce platforms has markedly heightened the frequency and intricacy of financial fraud. Conventional rule-based fraud detection systems frequently struggle to recognise advanced and changing fraudulent schemes because of their restricted adaptability and incapacity to handle extensive data sets. Machine Learning (ML) has surfaced as an effective method for recognising fraudulent transactions by autonomously discerning concealed patterns, recognising anomalies, and offering instantaneous decision assistance. This chapter offers an extensive summary of machine learning methodologies employed in fraud detection, encompassing supervised, unsupervised, semi-supervised, and reinforcement learning strategies. A range of methods, applications, obstacles, and prospective research avenues are examined within the framework of cloud-enabled IoT and e-commerce environments.
Dermatological disorders provide a considerable worldwide medical dilemma due to their varied clinical presentations, overlapping visual traits, and differing degrees of severity. Prompt and precise recognition of skin disorders is crucial for prompt intervention, suitable referral, and enhanced patient results. Conventional diagnosis of skin disorders predominantly depends on clinical assessment, dermoscopic analysis, patient medical history, and, where warranted, histological study. This chapter offers an extensive summary of AI-based skin disease identification, encompassing picture collection, preprocessing, feature extraction, classification, interpretable AI, model assessment, and clinical decision assistance. The section additionally addresses issues related to dataset imbalance, discrepancies in skin tones, insufficient data diversity, diagnostic ambiguity, privacy concerns, and clinical validation. The suggested paradigm establishes a basis for creating dependable, comprehensible, and fair AI-enhanced dermatological screening systems. Such systems ought to augment, rather than supplant, proficient clinical evaluation.
Retinal disorders rank among the primary contributors to vision loss and blindness globally. Disorders such diabetic retinopathy, glaucoma, age-related macular degeneration, hypertensive retinopathy, and retinal vascular irregularities can result in gradual vision deterioration if not recognised and treated well. This chapter offers an extensive examination of retinal image analysis utilising deep learning, encompassing image capture, preprocessing, public datasets, feature extraction, illness categorisation, lesion detection, explainable artificial intelligence, and clinical decision support. The section additionally addresses assessment metrics, dataset disparity, picture fidelity, model generalisability, patient confidentiality, and clinical verification. The suggested framework establishes a basis for creating dependable and comprehensible retinal screening technologies that can assist ophthalmologists and enhance access to first eye disease evaluations. Clinical implementation necessitates autonomous verification and suitable expert supervision.