2890 papers
This paper presents CGHC-S128, a hybrid encryption framework that combines a Complete Graph Hill Cipher with Shift128 residual encryption to securely and efficiently encrypt universal binary files of any size without padding, while ensuring mathematical key invertibility, perfect decryption, and strong resistance to cryptanalytic attacks.
This paper proposes the CAI Computing Power Law, a quantitative framework (C = A/I) demonstrating that optimizing algorithms and purifying information resources can significantly enhance effective computing power and mitigate diminishing returns, thereby guiding industries to achieve a dynamic tripartite balance among computing power, algorithms, and generalized information rather than relying solely on hardware expansion.
This paper proposes a big data-driven early warning model for tobacco enterprises that integrates a scalable distributed data pipeline, advanced feature engineering, and an ensemble learning system to overcome the limitations of traditional monitoring by enabling real-time, accurate, and interpretable safety risk detection and decision support.
The paper introduces OBFM-SafeEnsemble, a guarded cross-outbreak transfer learning framework that leverages source priors and adaptive regularization to significantly improve confirmation status prediction in data-scarce Ebola outbreaks while preventing harmful knowledge transfer through strict validation and ensemble selection.
This paper proposes an AI-driven adaptive hybrid cryptosystem that dynamically allocates Unimodular Hill Cipher and RSA-OAEP encryption based on file characteristics, utilizing Logistic Chaotic Maps for key generation to achieve robust, efficient, and lossless security for universal binary files.
This study presents a comprehensive bibliometric and systematic review of deep learning-based melanoma detection research from 2015 to 2026, highlighting key trends in CNN architectures and transfer learning while identifying critical challenges like dataset imbalance and clinical validation to guide future development of reliable diagnostic systems.
This paper introduces a confidence-gated multimodal framework that integrates text embeddings, acoustic features, and speech-guided emotion captions to dynamically weight modality reliability, achieving state-of-the-art depression detection performance while addressing evaluation leakage through rigorous nested cross-validation.
This paper reveals that TrustZone-M proofs of execution are vulnerable to state tampering via non-secure DMA controllers because their security guarantees rely solely on CPU isolation, and proposes a lightweight mitigation using per-channel DMA security to ensure state authenticity against all bus masters.
This study audits the impact of model compression techniques on subgroup fairness in ICU AKI risk prediction using MIMIC-IV, revealing that while naive quantization can catastrophically degrade performance and pruning significantly harms subgroup accuracy, proper methodological controls are essential to avoid spurious fairness claims and address inherent disparities driven by small subgroup sizes.
This study leverages MIMIC-IV data to demonstrate that while standard tabular models often outperform graph-based approaches for predicting acute kidney injury and bleeding, graph embeddings derived from drug co-administration networks provide statistically significant, interpretable improvements in predicting delirium risk when label leakage is rigorously corrected.