2608 papers
This paper presents OTNT, a self-destructing WireGuard tunnel scheme that enhances security by automatically terminating connections upon reaching either a time or data volume threshold, while maintaining low overhead and ensuring robust interface management through dynamic allocation and crash recovery.
This paper proposes a Caputo fractional differential equation model that couples residual energy, trust, and compromise probability to analyze how the fractional order's memory parameter influences the trade-off between security responsiveness and energy conservation in mobile-agent wireless sensor networks, demonstrating that while fractionalization alters transient dynamics under burst attacks, it does not intrinsically enhance security.
This study demonstrates that utilizing extended-reasoning inference significantly improves GPT-5's accuracy on Orthopaedic In-Training Examination questions compared to standard mode, though the persistence of confident errors suggests these models should serve as supervised adjuncts rather than primary study resources for residents.
This study presents a fine-tuned foundation model approach that leverages a newly curated dataset of 2,909 annotated frames to achieve high-accuracy semantic segmentation of surgical instruments and anatomy, as well as robust recognition of surgical action triplets, thereby advancing intraoperative guidance and automated skill assessment in gastric cancer surgery.
This study demonstrates that while short-term signal dynamics are crucial for predicting tics in Tourette syndrome, longer 5-second temporal windows combined with temporal CNNs significantly enhance the detection of motor tics from EMG data, highlighting the need for multimodal integration and patient-specific modeling in future research.
This study presents a robust multi-feature Transformer fusion framework that integrates patellar morphometry, radiomics, and deep learning features from CT scans to achieve highly accurate and generalizable sex estimation, outperforming single-feature models and forensic experts while significantly reducing processing time.
This working paper proposes and validates a seven-layer mathematical and computational framework for personalized Alzheimer's medication-adherence and cognitive-risk modeling, demonstrating through a synthetic cohort that temporal, participant-conditioned machine learning models significantly outperform population-level baselines in predicting adherence difficulties.
This paper introduces a certification framework for finite exact learning under bounded adversarial errors, utilizing isolation witnesses and portable certificates to prove optimal query complexities and demonstrate significant improvements in coverage and efficiency over non-adaptive strategies.
This paper presents an exact softmax sampling method for residual quantum overlaps that utilizes nested classical projections and a first-proposal coupling to significantly reduce the expected shot cost and variance, as demonstrated on pretrained model attention rows, though it does not establish a hardware speedup.
This paper proposes a hierarchical Spatial Basis Model that maps inputs to high-dimensional vector nodes and propagates information via cosine similarity-based inheritance and linear scaling, offering a computationally efficient alternative to MLPs for scenarios with pronounced spatial structures or high-dimensional discrimination needs.