Year: 2025
Type: article
Abstract: The rapid evolution of machine learning (ML) and artificial intelligence (AI) has led to groundbreaking advancements in computational models, empowering applications across diverse domains. This paper provides an in-depth exploration of advanced ML architectures, including transformers, Graph Neural Networks (GNNs), capsule networks, spiking neural networks (SNNs), and hybrid models. These architectures address the limitations of traditional models like convolutional and recurrent neural networks, offering superior accuracy, scalability, and efficiency for complex data. Key applications are discussed, ranging from healthcare diagnostics and drug discovery to financial fraud detection, autonomous systems, and logistics optimization. Despite their potential, these architectures face challenges such as computational overhead, scalability, and interpretability, necessitating interdisciplinary solutions. The paper also outlines future directions in edge computing, explainable AI, quantum machine learning, and few-shot learning, emphasizing the transformative role of advanced ML architectures in reshaping AI’s future.
DOI: 10.14419/kwb24564
Authors: Ridwan Boya Marqas, Saman M. Almufty, Renas R. Asaad, Dr. Tamara Saad mohamed
Institutions: Knowledge University, University of Kut
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W4407808676