feature-selection

The paper proposes 'predictive closure,' a set-level characterization of how JEPA selects features to encode, showing that feature retention depends on relationships within feature sets rather than individual predictability, with experiments demonstrating that unpredictable features can be preserved when they aid prediction of others.

该论文提出对特征选择层（Feature Selection Layer）的多项扩展，用于神经网络的嵌入式特征加权，以提升多分类任务中标签特定相关性的捕获能力，并在基因表达数据等基准上优于现有方法。

This paper evaluates 21 Natural Visibility Graph topological metrics for cyber-attack classification using SHAP, Permutation Importance, Boruta, and RFE consensus ranking, showing that a Top3 clustering-coefficient subset achieves 97.15% accuracy on CICIDS2018 while cutting runtime by 96%.

The paper shows that statistical top-k feature selection for SAE-based steering of LLMs is suboptimal and proposes Neighbor Integrated Feature Selection (NIFS) to improve performance by leveraging representation similarity.

This study compares mutual information and data-based sensitivity analysis for feature selection in bank telemarketing, showing that sensitivity analysis achieves good prediction with fewer features while mutual information is better for cost reduction.

The paper introduces PAFIR, an adaptive feature selection framework using reinforcement learning for personalized fall risk prevention from longitudinal multimodal health data, demonstrating improved effectiveness over baseline methods.

This paper proposes a new asymmetric robust bounded sparse smooth (aR) loss function for l1-norm penalized geometric twin support vector machine (aRSGTSVM) to handle classification and regression tasks with label and feature noise, achieving feature selection and robustness. Experiments on synthetic and UCI datasets plus China stock market index tracking demonstrate superiority.

This paper compares five feature selection methods for EHR diagnosis codes in opioid use disorder prediction, finding that NTK sensitivity offers the best accuracy-stability balance while LLM-guided selection adds complementary clinical signals.

This paper introduces PopFS, a method for learning a shared feature set that is robust across heterogeneous populations, allowing each population to train its own model. Experiments on diverse tabular datasets and a 43-state COVID-19 nowcasting study show strong average and worst-population performance.

This arXiv paper proposes a three-class detection framework distinguishing humans, bots, and AI agents, showing binary classifiers systematically misclassify agents. It identifies minimal feature sets (e.g., mouse_event_rate and teleport_click_ratio) that achieve perfect agent recall across evasion levels.

This paper proposes a k-order relaxation of the faithfulness assumption for learning graphical Markov blankets, and introduces a proof-of-concept algorithm (kOMB) that can recover Markov blankets even under violations of faithfulness, such as parity-type relationships.

This paper presents an Automated Data Processing (ADP) framework that uses reinforcement learning and Shapley-based explainable AI to optimize machine learning model-feature combinations for warpage detection in fused deposition modeling, achieving improved accuracy and stability.

This paper introduces drXAI, a method that uses XAI attribution to reduce data size for time series classification, achieving 80-90% data reduction while maintaining accuracy, enabling large models to scale.

This study proposes a feature-guided zero-shot framework using LLMs for early chronic kidney disease screening, achieving consistent improvements with minimal community-accessible features across heterogeneous datasets.

This paper presents a granularity-aware EEG feature framework that organizes multi-scale descriptors into global, regional, and channel levels to predict dimensional psychopathology. Using the HBN cohort, it shows that tree-based models and granularity-balanced feature selection yield modest improvements, suggesting multi-scale EEG features contain weak but detectable signals for pediatric mental health.

This paper studies conditional inference trees and forests for feature selection, showing that CIF ranks 4th among 17 classification methods and 3rd among 18 regression methods. Runtime ablations reveal adaptive stopping and threshold search have the largest effect on runtime with minimal impact on downstream scores.

This paper proposes a hybrid predictive model combining ensemble feature selection (ANOVA and mutual information) with Harris Hawks optimization-tuned logistic regression for explainable mental health risk prediction in female sex workers, achieving 95.78% accuracy.

This paper introduces Multi-Adapter PPO, a reinforcement learning framework with cross-attention for wavelength selection in LIBS quantitative analysis, achieving 28.4% better composite scores and 45.2% improvement in prediction accuracy over traditional methods on steel and coal datasets.

This paper introduces nCMD, a benign-anchored feature selection method for imbalanced network intrusion detection that scores features by deviation from benign class mean, outperforming classical filters on benchmark datasets.

This paper investigates whether real-world datasets contain natural experiments by using causal discovery and feature selection, finding that they do and can improve model performance.