Feature selection plays a pivotal role in enhancing machine learning models by mitigating dimensionality and reducing computational overhead. This survey presents a comprehensive examination of recent advancements in feature selection techniques, with a particular emphasis on meta-heuristic algorithms. Renowned for their adaptability and efficacy, meta-heuristics have emerged as powerful tools for addressing the combinatorial complexities inherent in feature selection. The study systematically categorizes and analyzes prominent meta-heuristic approaches, including evolutionary algorithms, swarm intelligence, and hybrid methodologies, while critically assessing their advantages, drawbacks, and practical applications across diverse domains. Furthermore, the paper explores the synergistic integration of meta-heuristics with conventional optimization techniques and machine learning frameworks, shedding light on prevailing trends and unresolved challenges. Finally, the discussion outlines promising future research directions, underscoring the potential of meta-heuristic-driven feature selection in managing high-dimensional datasets and solving intricate real-world problems.