In high-dimensional data, features often exhibit complex correlations and redundancies that hinder effective learning and reduce model interpretability. Therefore, extracting critical feature structural information from such data is essential for effective feature selection. This structural information can be represented as a feature graph, which reveals associations among features, including redundancies and correlations. However, most existing feature graph-based methods adopt a single strategy, focusing either on selecting relevant features or removing redundant ones, often resulting in suboptimal performance. Moreover, these methods typically construct static feature graphs using all input features, which may introduce irrelevant or detrimental relationships. Such simplistic approaches and low-quality graphs limit the effectiveness of feature selection. To address these limitations, we propose the Dynamic Feature Graph (DFG) framework that jointly learns feature graphs and selects features. By decoupling feature associations according to category, the DFG framework dynamically identifies key feature relationships, eliminating redundancy across classes while preserving essential intra-class correlations. This results in the selection of both relevant and non-redundant features. Additionally, we introduce sample manifolds with a rank equality constraint to ensure the selection of category-related features. To the best of our knowledge, this is the first dynamic feature graph approach in the field. Extensive experiments on 15 widely used real-world datasets demonstrate that DFG outperforms 20 state-of-the-art feature selection methods, highlighting its effectiveness in supervised and unsupervised settings.
Understanding feature associations is vital for selecting the most informative features. Existing methods primarily focus on global feature associations, which capture overall relationships across all samples. However, they often overlook class-specific feature interactions, which are essential for capturing locality features that may only be significant within certain classes. In this paper, we propose Dual-Regularized Feature Selection (DRFS), which incorporates two feature association regularizers to address both class-specific and global feature relationships. The class-specific regularizer captures the local geometric structure of features within each class. Meanwhile, the global regularizer utilizes a global feature similarity matrix to eliminate redundant features across classes. By combining these regularizers, DRFS selects features that preserve both local interactions within each class and global discriminative power, with each regularizer complementing the other to enhance feature selection. Experimental results on eight public real-world datasets demonstrate that DRFS outperforms existing methods in classification accuracy.