2661 papers
This paper presents a GPU-accelerated pipeline that maps integer factorization to a sparse Ising model solved via graph-colored simulated annealing on an NVIDIA GH200, successfully factoring 128-bit semiprimes by combining parallel spin updates with guided post-processing techniques.
This study demonstrates that while transfer learning significantly enhances the performance of pretrained GeoAI models for rooftop solar PV mapping across geographic regions, predictive accuracy alone is insufficient to guarantee trustworthy deployment, necessitating a comprehensive evaluation framework that also accounts for robustness, reliability, generalization, transparency, and residual uncertainty.
This paper introduces AAPS-TLS, an open-source, multi-agent AI framework that leverages LLM-driven reasoning and reinforcement learning to achieve near-perfect predictive certificate management, encrypted traffic anomaly detection, and automated policy compliance for SSL/TLS infrastructure in cloud-native environments, validated through rigorous reproducible simulations and indicative production pilots.
This paper presents ALJ, a hybrid two-tier continuous authentication framework for Zero-Trust web deployments that replaces proxy models with genuine ARIMA, LCS, SVM, and JRip algorithms and introduces a security-first calibration strategy to significantly improve recall and reduce adversary bypass rates compared to standard static thresholds.
This paper proposes FDC-DETR, a lightweight and enhanced RT-DETR framework that integrates Fourier convolution, dynamic entropy-aware attention, and context-refined fusion to achieve high-accuracy, real-time detection of pulmonary nodules in CT images while effectively suppressing background interference.
This study developed a two-stage workflow combining an improved YOLOv8s machine vision model for initial leafhopper damage screening with a near-infrared spectroscopy model for reassessing uncertain cases, achieving a 89.5% overall grading accuracy for fresh tea leaves that significantly outperforms standalone visual assessment.
This paper presents an end-to-end framework that enables quadruped robots to execute free-form natural language commands by combining a lightweight NLU module with a goal-conditioned PPO policy and teacher-student distillation, achieving robust, adaptive, and energy-efficient locomotion in both simulation and real-world experiments.
The paper proposes Steel-CGA, a lightweight YOLO26s-based detector that integrates Channel-Spatial-Semantic Attention, Grouped Dilated Residual Fusion, and Adaptive Downsampling modules to significantly improve real-time steel surface defect detection accuracy while reducing computational costs.
This paper demonstrates that in complex networks, strategically dispersing multiple initial rumour spreaders to minimize spatial overlap and delay deactivation significantly enhances both the final reach and peak intensity of rumour propagation compared to traditional degree-based seeding strategies.
This paper introduces instruction graph scheduling within the Celerity runtime to move complex memory and communication analysis out of the critical path for SYCL programs on multi-GPU systems, enabling concurrent execution and optimized memory allocation that achieves strong scaling across up to 128 GPUs.