<p>Industry 4.0 has redefined modern manufacturing by integrating cyber&ndash;physical systems, Industrial Internet of Things (IIoT), cloud&ndash;edge computing, and data-driven intelligence. Among these enablers, machine learning (ML) has emerged as a foundational technology for extracting actionable insights from heterogeneous manufacturing data. This paper presents an extended and comparative review of ML and deep learning (DL) techniques&mdash;including supervised, unsupervised, semi-supervised, reinforcement learning, and hybrid models&mdash;applied across core manufacturing domains such as predictive maintenance, quality inspection and defect detection, process optimization, production planning, and supply chain management. Based on a systematic analysis of literature published between 2015 and 2025, the review compares algorithmic performance, computational complexity, interpretability, and deployment feasibility. Mathematical formulations of commonly used models, including regression, support vector machines, convolutional neural networks (CNNs), and long short-term memory (LSTM) networks, are presented to enhance methodological clarity. Emerging trends such as transfer learning, federated learning, edge AI, and explainable artificial intelligence (XAI) are discussed in the context of industrial scalability and reliability. The study concludes that context-aware model selection, combined with hybrid and explainable frameworks, is critical for bridging the gap between laboratory-scale ML models and real-world smart manufacturing systems.</p>