2731 papers
This study demonstrates that an Adaptive Gated DenseNet121 model improves paddy disease classification accuracy over a standard baseline and enables limited parameter pruning, though it currently falls short of achieving robust mobile compression due to performance collapse at higher sparsity thresholds.
This paper presents a systematic review of deep learning, federated learning, and meta-learning approaches for IoT intrusion detection, classifying existing works across six key dimensions while identifying critical limitations such as non-comparable evaluation protocols, insufficient handling of non-IID data, and a lack of reproducibility in code and edge measurements.
This paper proposes a CLIP-based framework that mitigates source-domain dependency in target-free zero-shot medical anomaly detection through multi-level strategies including prompt learning, adversarial visual adaptation, and feature consistency, thereby achieving robust generalization across heterogeneous medical imaging datasets without access to target data.
Scene-Guard introduces a Scene Structural Disruption Module (SSDM) that successfully preserves measurable scene-manipulation disruption across multiple editors and JPEG recompression levels, though it is currently limited to fixed-mask scenarios and prioritizes structural divergence over imperceptibility.
This paper proposes a hybrid framework for long-distance Quantum Key Distribution that combines an SVM-RBF classifier using QBER, Bell-CHSH, and Temporal Shannon Entropy features to detect security breaches with 87.82% accuracy, and a chaotic perturbation method enhanced by SHA-256 whitening to generate NIST-compliant secure session keys.
This paper proposes an online reviews-driven large-scale group decision-making framework that integrates social network analysis, particle swarm optimization, and fuzzy rough set models to effectively analyze consumer preferences for fresh fruit on e-commerce platforms and support the development of intelligent shopping assistants.
This paper proposes ARGO-5G, a leakage-aware robust submodular optimization algorithm that prioritizes 5G infill sites by balancing future demand capture, regional service equity, and engineering constraints without relying on future labels, demonstrating superior portfolio balance and deployability on a dataset of over 14,000 macro sites.
The paper introduces SIR-PhysNet, an ultra-lightweight, deadline-aware rPPG pipeline that offloads geometric and illumination processing to fixed-cost operations and uses an analytical Fourier mask for heart rate extraction, enabling real-time, high-accuracy monitoring on resource-constrained edge devices like the Raspberry Pi 5 where face detection, rather than network inference, is the primary performance bottleneck.
AdaFuse is a cross-modal image fusion algorithm that addresses feature imbalance and computational inefficiency by integrating negative value calibration for precise attention selection with an adaptive hierarchical activation strategy for dynamic resource allocation, achieving superior fusion quality and efficiency in visible-infrared image fusion tasks.
This study proposes a quantitative Digital Twin Auditability Framework (DTAI) that integrates eight audit dimensions and a residual risk model to assess the trustworthiness of enterprise decision systems, revealing through empirical evaluation that while current environments are generally controlled, significant weaknesses in business continuity, traceability, and model governance require targeted improvements to mitigate risks like unauthorized access and data manipulation.