This chapter discusses both global and local methods for feature selection, providing a brief overview of evolutionary algorithms and neural networks in computational intelligence. Global feature selection is currently the most widely applied method, leveraging its advantages such as global search capabilities, high adaptability, flexibility, and the ability to perform multi-objective optimization. Local feature selection introduces a novel approach, optimizing within local feature spaces, offering a convenient way to handle heterogeneous structured data and, to some extent, achieving better efficiency. Evolutionary algorithms, a category simulating biological evolution, find extensive applications in addressing complex optimization and search problems due to their global search capabilities and adaptability to intricate issues. Deep learning, characterized by constructing deep neural networks, facilitates automatic learning of abstract representations of complex data patterns. This enables the model to achieve significant performance improvements when dealing with large-scale datasets and complex tasks.