Describe advanced feature selection techniques and their trade-offs.

Advanced feature selection combines statistical methods, machine learning algorithms, and domain expertise for optimal feature subsets.\n\n• Methods: Recursive feature elimination, LASSO regularization, mutual information\n• Wrapper methods: Forward/backward selection with cross-validation\n• Embedded methods: Tree-based importance, neural network attention\n• Evaluation: Stability analysis, performance vs complexity trade-offs\n\nExample: High-dimensional genomics data uses LASSO for sparse feature selection, validates stability across bootstrap samples, combines with biological pathway knowledge, and evaluates predictive performance using nested cross-validation.