Title: Deep Learning for Big Data in Supervised Learning: A Systematic Review of Architectures, Methods, Applications and Future Directions
Authors: Virendra Tank
Volume: 10
Issue: 7
Pages: 163-170
Publication Date: 2026/07/28
Abstract:
The exponential growth of digitally generated information has positioned big data as both an unprecedented opportunity and a formidable analytical challenge for modern computing systems. Deep learning, a subfield of machine learning characterized by hierarchical, multi-layered representation learning, has emerged as the dominant paradigm for extracting actionable knowledge from massive, high-dimensional, and heterogeneous datasets. When coupled with supervised learning - in which models are trained on labeled examples to approximate an input-output mapping - deep architectures such as convolutional neural networks, recurrent neural networks, and transformers have achieved state-of-the-art performance across computer vision, natural language processing, healthcare analytics, and industrial monitoring. This paper presents a systematic review of the intersection between deep learning, big data, and supervised learning. Following a PRISMA-inspired methodology, 27 peer-reviewed studies published between 1997 and 2025 were analyzed and thematically synthesized. The review develops a taxonomy of supervised deep learning architectures applied to big data, compares their reported performance and computational trade-offs, and identifies four persistent challenges: scalability, label scarcity, interpretability, and data veracity. The paper concludes with a discussion of emerging directions, including self-supervised pretraining, federated and distributed deep learning, and energy-efficient model compression, that are expected to shape the next generation of big-data-driven supervised learning systems.