2890 papers
This corpus-based study demonstrates that a regression model integrating both lexical and syntactic complexity metrics effectively predicts machine translation quality in German-to-Chinese philosophical texts, highlighting the critical role of holistic linguistic features in assessing translation outcomes.
This paper presents a lightweight, modular framework that integrates monocular dense SLAM (MASt3R-SLAM) with vision-language models (Gemma 3 and Qwen2.5-VL) to enable robust, open-vocabulary semantic navigation in indoor environments using only RGB input, without requiring depth sensors or model fine-tuning.
This paper introduces XAI-Gate, a novel, detector-agnostic governance framework that dynamically selects explanation actions to balance packet quality of service, operator interpretability, and information exposure under resource constraints, achieving significant reductions in explanation debt and zero budget violations without requiring classifier re-training.
This paper introduces Omnilingual ASR, a scalable, open-source family of speech recognition models that leverages a 7B-parameter self-supervised architecture and a massive, diverse training corpus to support over 1,660 languages, including more than 500 previously unserved, while enabling communities to easily adapt the system to new languages with minimal data.
This paper introduces UDAQ, a unified, reward-driven Q-iteration framework that autonomously adapts to dynamically evolving state spaces to simultaneously optimize both point-to-point path planning and efficient environment exploration, significantly outperforming conventional planners and exploration strategies in unknown environments without prior map knowledge.
This paper proposes a novel pose-robust finger identification framework that integrates mutual region alignment, minimum convolution point feature learning, and weighted score fusion to effectively address performance degradation caused by longitudinal rotation and non-rigid deformations in contactless finger images, achieving state-of-the-art results across four public datasets.
This paper presents a behavior-centric framework for classifying MITRE ATT&CK techniques using Sysmon telemetry, demonstrating that GUID-based event correlation combined with the SecureBERT transformer model outperforms temporal correlation and traditional machine learning approaches in identifying adversarial activities.
This paper proposes a novel Rectified Sine Unit (RSU) activation function that significantly enhances the performance of Convolutional Neural Networks in automated pathological diagnosis across four diverse cancer datasets, achieving superior accuracy compared to standard activation functions like ReLU and Leaky ReLU.
This study empirically validates a hybrid AI-IoT framework that integrates continuous environmental sensing with machine learning to predict and provide early warnings for chromatic deterioration on tropical heritage façades in Indonesia, achieving high accuracy and generalizability across multiple sites.
This paper proposes a deep learning-based security state awareness method for petrochemical industrial control network equipment using a hybrid TCN-MGRU model to address situational complexity and information delays, thereby significantly improving detection accuracy and security protection.