When Features Become Instances: Inverted Contrastive Learning for Unsupervised Feature Selection
本文提出了一种基于特征一致性而非样本一致性的无监督特征选择方法ICLFS,通过构建正负视图并使用InfoNCE目标学习投影空间中的表示来解决特征选择问题。
本文提出了一种基于特征一致性而非样本一致性的无监督特征选择方法ICLFS,通过构建正负视图并使用InfoNCE目标学习投影空间中的表示来解决特征选择问题。
本文提出MADS,一种19维物理信息描述符集,用于捕捉音频信号中的多种结构,并在多个数据集上证明了其有效性。
This study addresses the challenges of demand uncertainty, variable replenishment lead times, and limited shelf life in pharmaceutical supply chains by formulating dynamic inventory management as a Markov decision process. The authors propose a hybrid deep reinforcement learning algorithm that integrates asynchronous advantage actor-critic (A3C) with distributed proximal policy optimization (DPPO) to dynamically optimize replenishment policies in continuous action spaces. This approach effectively tackles sequential decision-making under stochastic conditions, achieving high patient service levels while substantially reducing inventory costs. Experimental results based on real-world pharmaceutical inventory data demonstrate that the proposed method consistently outperforms benchmark approaches across diverse dynamic scenarios, highlighting its practicality and effectiveness.
This study addresses the challenge of detecting multimodal fake news in India’s diverse media landscape by proposing a novel approach that integrates visual and textual modalities. The method leverages ResNet-50 to extract image features and DistilBERT to obtain semantic embeddings, and uniquely introduces an Adaptive Neuro-Fuzzy Inference System (ANFIS) into multimodal fake news detection to better model ambiguity and uncertainty. A lightweight, learnable-weight attention-based fusion module is further designed to dynamically integrate multimodal representations. Experimental results on the IFND dataset demonstrate that the proposed framework significantly outperforms existing methods across key evaluation metrics, including accuracy, precision, recall, and F1 score.
This work addresses the lack of concept-level interpretability and imbalanced concept attribution—leading to misclassifications—in toxic language detection. Methodologically: (1) it treats semantic subtypes (e.g., insult, threat, identity attack) as interpretable concepts, constructs a target lexicon, and proposes a Word–Concept Alignment (WCA) score to quantify each token’s contribution to misclassification via concept gradients (CG); (2) it introduces, for the first time, a delexicalized generative data augmentation strategy to assess model reliance on abstract toxic patterns rather than surface lexical cues. Experiments demonstrate that CG precisely identifies critical toxic tokens and reveal that models over-attribute toxicity to conceptual features even when explicit toxic words are absent—exposing systematic generalization biases toward deep semantic patterns and implicit dependency mechanisms. This establishes a novel paradigm for interpretable toxic language detection grounded in concept-level attribution and causal probing.
本文提出了一种基于特征一致性而非样本一致性的无监督特征选择方法ICLFS,通过构建正负视图并使用InfoNCE目标学习投影空间中的表示来解决特征选择问题。
本文提出MADS,一种19维物理信息描述符集,用于捕捉音频信号中的多种结构,并在多个数据集上证明了其有效性。
This study addresses the challenges of demand uncertainty, variable replenishment lead times, and limited shelf life in pharmaceutical supply chains by formulating dynamic inventory management as a Markov decision process. The authors propose a hybrid deep reinforcement learning algorithm that integrates asynchronous advantage actor-critic (A3C) with distributed proximal policy optimization (DPPO) to dynamically optimize replenishment policies in continuous action spaces. This approach effectively tackles sequential decision-making under stochastic conditions, achieving high patient service levels while substantially reducing inventory costs. Experimental results based on real-world pharmaceutical inventory data demonstrate that the proposed method consistently outperforms benchmark approaches across diverse dynamic scenarios, highlighting its practicality and effectiveness.
This study addresses the challenge of detecting multimodal fake news in India’s diverse media landscape by proposing a novel approach that integrates visual and textual modalities. The method leverages ResNet-50 to extract image features and DistilBERT to obtain semantic embeddings, and uniquely introduces an Adaptive Neuro-Fuzzy Inference System (ANFIS) into multimodal fake news detection to better model ambiguity and uncertainty. A lightweight, learnable-weight attention-based fusion module is further designed to dynamically integrate multimodal representations. Experimental results on the IFND dataset demonstrate that the proposed framework significantly outperforms existing methods across key evaluation metrics, including accuracy, precision, recall, and F1 score.
This work addresses the lack of concept-level interpretability and imbalanced concept attribution—leading to misclassifications—in toxic language detection. Methodologically: (1) it treats semantic subtypes (e.g., insult, threat, identity attack) as interpretable concepts, constructs a target lexicon, and proposes a Word–Concept Alignment (WCA) score to quantify each token’s contribution to misclassification via concept gradients (CG); (2) it introduces, for the first time, a delexicalized generative data augmentation strategy to assess model reliance on abstract toxic patterns rather than surface lexical cues. Experiments demonstrate that CG precisely identifies critical toxic tokens and reveal that models over-attribute toxicity to conceptual features even when explicit toxic words are absent—exposing systematic generalization biases toward deep semantic patterns and implicit dependency mechanisms. This establishes a novel paradigm for interpretable toxic language detection grounded in concept-level attribution and causal probing.