2787 papers
The paper proposes RAUDI, a dual-head framework for industrial anomaly detection that unifies unsupervised and supervised learning by flexibly utilizing real defective samples or realistic pseudo-anomalies synthesized via a novel multi-stage superpixel method (RASMS), achieving state-of-the-art performance on both MVTec AD and KolektorSDD2 benchmarks.
This paper audits the ACL 2026 proceedings to find that while nearly all cited references exist, a single-run automated audit of claim support failed to replicate due to an unlogged model configuration, demonstrating that rigorous citation verification requires strict operational logging to ensure reproducibility.
This paper examines the integration of AI and ML into cybersecurity operations, highlighting their potential to enhance threat detection and efficiency while emphasizing that successful deployment requires addressing challenges like adversarial attacks and data quality through robust governance, transparency, and oversight.
This paper proposes an explainable AI-based hybrid Intrusion Detection System for IoT and computer networks that integrates ADASYN for data imbalance, filter methods for feature selection, ensemble learning for robustness, and SHAP for interpretability, demonstrating superior performance on CICIDS2017 and MQTT-IoT-IDS2020 datasets.
This paper proposes a weakly supervised framework that leverages image-level annotations to generate refined pseudo-masks via CAM and Grad-CAM, which are then used to train a hybrid OUR-Net–U-Net model for accurate, multi-domain structural crack segmentation with significantly reduced annotation costs and superior performance compared to existing methods.
This study analyzes 849 AI incidents to empirically derive three distinct harm profiles—Autonomous/Embodied, Vision Recognition, and Generative AI—arguing for a shift from broad governance principles to a context-sensitive framework that tailors accountability and monitoring mechanisms to these specific risk patterns.
This paper proposes a defect-aware YOLOv8n framework enhanced with a ResNet50 backbone, DeCA, improved BiFPN with SPD-Conv, a dynamic adaptive decoupled head, and Wise-IoU v3 to achieve real-time, high-accuracy detection of small, low-contrast defects on printed playing cards amidst dense graphic backgrounds, outperforming the standard YOLOv8n by 12.8% in mAP@50 while maintaining a 97 FPS inference speed.
This paper proposes Prototype-aware Channel-Selective Spatial Interaction (PCSI), a teacher-student framework for semi-supervised medical image segmentation that leverages decoupled foreground-background prototype memories to guide channel selection and spatial feature refinement, thereby achieving superior performance with limited labeled data.
This paper introduces EMA-Det, a hardware-aware, early-exit region-verification network for two-stage UAV detection that prioritizes recall over mAP to overcome system bottlenecks, achieving superior performance on edge devices and demonstrating the necessity of domain-specific training for thermal-IR applications.
This paper proposes Semantic Spectrum-based Fault Localization (SSFL), a method-level approach that leverages runtime output-value distributions to construct semantic spectra, achieving superior fault localization accuracy compared to traditional spectrum-based, learning-based, and LLM-based techniques without requiring model training or online reasoning.