2787 papers
This paper presents a comprehensive survey that systematically reviews research on detecting user-generated toxic content on Chinese social media and safeguarding Chinese large language models against LLM-generated toxicity, aiming to consolidate fragmented progress and identify key challenges for future development.
The paper proposes EPDA, an efficient and privacy-preserving dual aggregation scheme for Federated Learning in VANETs that ensures model confidentiality, anonymity, and unlinkability while significantly reducing computational and communication overheads.
This paper proposes a cost-effective, robust two-way authentication protocol for low-cost RFID systems in IoT environments that integrates the Rabin cryptosystem with random number generation to enhance privacy, prevent replay and brute force attacks, and ensure information freshness.
This paper proposes a parameter-efficient Parkinson's disease recognition framework that leverages pre-trained visual foundation models with three lightweight adapters (global, local, and token-wise) to achieve competitive performance on OCT images while significantly reducing trainable parameters compared to traditional deep neural networks.
This paper proposes a three-stage, concavity-aware harmonic field–point–line framework that achieves high-accuracy and efficient 3D tooth segmentation for orthodontic CAD by combining adaptive ellipsoidal cropping, a Locational Strength metric, and boundary-aware graph search, demonstrating robust clinical applicability on the Teeth3DS benchmark.
This study benchmarks nine machine learning models on Dutch motorway data to predict five types of unsafe road conditions, finding that Random Forest and XGBoost outperform deep learning approaches while demonstrating that weather features are more predictive than infrastructure proximity, though data sparsity and coarse features currently limit the models' reliability for calibrated risk forecasting.
JANUS is a high-performance architecture that federates Trusted Execution Environments across untrusted networks using a split Trusted Computing Base design, achieving native-speed encrypted routing with minimal latency overhead by anchoring key release to hardware measurements and executing enforcement within the kernel.
SimGrowNet is an efficient neural growth method inspired by human neurogenesis that progressively expands compact models during training while minimizing neuron redundancy through similarity-based constraints, thereby enhancing feature diversity, mitigating shortcut learning, and outperforming existing baselines on multiple image recognition datasets.
This study demonstrates that while WordBinary's text similarity detection performance systematically declines as reused content is fragmented into shorter passages, the system remains robust and effective, successfully recovering over 90% of deliberately reused words even when dispersed across numerous small fragments.
This paper proposes a fuzzy graph-based methodology to model uncertainty in social networks by redefining classical centrality measures, demonstrating through a case study of University of Tehran faculty that this approach captures nuanced relationship strengths and yields distinct actor rankings compared to traditional crisp models.