A Smallest-Need-First Job Scheduling Framework with Adaptive Optimization of Idle Node Counts for Energy-Efficient HPC Systems
本文提出SNF-ICON框架，通过结合最小需求优先调度、预测唤醒时间和自适应温备控制来优化高性能计算系统的能源效率，减少空闲能耗和唤醒延迟。
0 citationsRead paper
Selected work
Representative Papers
Beyond Local Power: Functional Connectivity Analysis for Subject-Independent Learning Style Recognition
This study addresses the limitations of traditional questionnaires and behavioral logs in learning style identification—namely, their subjectivity or reliance on prolonged data collection—by introducing functional brain connectivity for cross-subject learning style recognition. Using EEG-derived phase-locking values (PLV) to construct inter-regional brain connectivity, the authors employ support vector machines with leave-one-subject-out cross-validation to classify learners along the active–reflective and verbal–visual dimensions of the Felder–Silverman model. Results reveal that fronto-occipital polarization features are discriminative for the verbal–visual dimension, achieving 70.00% subject-level accuracy, whereas performance on the active–reflective dimension remains modest at 55.56%, likely due to overlapping executive network engagement and systematic inter-individual differences in neural connectivity. Notably, the findings uncover a “systematic neural inversion” phenomenon, challenging the assumption of a universal classifier across subjects.
0 citationsRead paper
Aligning Implied Statements for Implicit Hate Speech Generalizability with Context-Bounded Semi-hard Negative Mining
This work addresses the challenge of implicit hate speech detection, where models often exhibit poor generalization due to reliance on contextual innuendo rather than explicit expressions. The authors propose ImpSH, a novel framework that uniquely integrates implicit statement alignment with context-constrained semi-hard negative mining. By leveraging triplet learning to focus on confusable instances and incorporating data augmentation to construct positive pairs—yielding the AugSH variant—the approach combines supervised contrastive learning with representation alignment on top of BERT or HateBERT. This effectively mitigates overfitting to superficial lexical cues. Experimental results demonstrate that ImpSH significantly outperforms baseline methods on the IHC, SBIC, and DynaHate datasets, achieving superior cross-domain detection performance, tighter intra-class representations, and more balanced global feature distributions.
0 citationsRead paper
Lius: Translation Model Based Instructional Lingustic Using Continual Instruction Tuning In Kupang Malay
This study addresses the significant performance degradation of large language models in translating low-resource languages—such as Kupang Malay—due to the scarcity of parallel corpora. To mitigate reliance on extensive parallel data, the authors propose a novel approach that integrates bilingual dictionary–driven explicit lexical and semantic feature instructions with continual instruction tuning (CIT), marking the first fusion of these two strategies. The resulting model, Lius, achieves consistent improvements of 4–6 points over standard instruction tuning across multiple automatic evaluation metrics and outperforms both state-of-the-art neural machine translation systems and multilingual large language models by margins of 10–13 points, substantially enhancing translation quality for low-resource languages.
0 citationsRead paper
From Patterns to Policy: A Scoping Review Based on Bibliometric Analysis (ScoRBA) of Intelligent and Secure Smart Hospital Ecosystems
This study addresses the fragmented state of research on intelligence and security in smart hospitals, which has lacked a systematic review and actionable policy guidance. Integrating the PAGER framework with a bibliometric approach (ScoRBA), the authors analyze 891 Scopus-indexed journal articles using co-occurrence analysis, network visualization, overlay analysis, and Enhanced Strategic Diagrams (ESD). They identify three core research clusters—artificial intelligence, privacy preservation, and cloud-edge architectures—and reveal critical gaps, particularly in interoperability and cross-layer integration. The work highlights emerging directions such as explainable AI and federated learning, and proposes an evidence-based policy pathway tailored for developing countries, centered on synergistic governance, scalable infrastructure, and secure data ecosystems.
0 citationsRead paper
Recent publications
Latest Papers
A Smallest-Need-First Job Scheduling Framework with Adaptive Optimization of Idle Node Counts for Energy-Efficient HPC Systems
本文提出SNF-ICON框架，通过结合最小需求优先调度、预测唤醒时间和自适应温备控制来优化高性能计算系统的能源效率，减少空闲能耗和唤醒延迟。
0 citationsRead paper
Beyond Local Power: Functional Connectivity Analysis for Subject-Independent Learning Style Recognition
This study addresses the limitations of traditional questionnaires and behavioral logs in learning style identification—namely, their subjectivity or reliance on prolonged data collection—by introducing functional brain connectivity for cross-subject learning style recognition. Using EEG-derived phase-locking values (PLV) to construct inter-regional brain connectivity, the authors employ support vector machines with leave-one-subject-out cross-validation to classify learners along the active–reflective and verbal–visual dimensions of the Felder–Silverman model. Results reveal that fronto-occipital polarization features are discriminative for the verbal–visual dimension, achieving 70.00% subject-level accuracy, whereas performance on the active–reflective dimension remains modest at 55.56%, likely due to overlapping executive network engagement and systematic inter-individual differences in neural connectivity. Notably, the findings uncover a “systematic neural inversion” phenomenon, challenging the assumption of a universal classifier across subjects.
0 citationsRead paper
Aligning Implied Statements for Implicit Hate Speech Generalizability with Context-Bounded Semi-hard Negative Mining
This work addresses the challenge of implicit hate speech detection, where models often exhibit poor generalization due to reliance on contextual innuendo rather than explicit expressions. The authors propose ImpSH, a novel framework that uniquely integrates implicit statement alignment with context-constrained semi-hard negative mining. By leveraging triplet learning to focus on confusable instances and incorporating data augmentation to construct positive pairs—yielding the AugSH variant—the approach combines supervised contrastive learning with representation alignment on top of BERT or HateBERT. This effectively mitigates overfitting to superficial lexical cues. Experimental results demonstrate that ImpSH significantly outperforms baseline methods on the IHC, SBIC, and DynaHate datasets, achieving superior cross-domain detection performance, tighter intra-class representations, and more balanced global feature distributions.
0 citationsRead paper
Lius: Translation Model Based Instructional Lingustic Using Continual Instruction Tuning In Kupang Malay
This study addresses the significant performance degradation of large language models in translating low-resource languages—such as Kupang Malay—due to the scarcity of parallel corpora. To mitigate reliance on extensive parallel data, the authors propose a novel approach that integrates bilingual dictionary–driven explicit lexical and semantic feature instructions with continual instruction tuning (CIT), marking the first fusion of these two strategies. The resulting model, Lius, achieves consistent improvements of 4–6 points over standard instruction tuning across multiple automatic evaluation metrics and outperforms both state-of-the-art neural machine translation systems and multilingual large language models by margins of 10–13 points, substantially enhancing translation quality for low-resource languages.
0 citationsRead paper
From Patterns to Policy: A Scoping Review Based on Bibliometric Analysis (ScoRBA) of Intelligent and Secure Smart Hospital Ecosystems
This study addresses the fragmented state of research on intelligence and security in smart hospitals, which has lacked a systematic review and actionable policy guidance. Integrating the PAGER framework with a bibliometric approach (ScoRBA), the authors analyze 891 Scopus-indexed journal articles using co-occurrence analysis, network visualization, overlay analysis, and Enhanced Strategic Diagrams (ESD). They identify three core research clusters—artificial intelligence, privacy preservation, and cloud-edge architectures—and reveal critical gaps, particularly in interoperability and cross-layer integration. The work highlights emerging directions such as explainable AI and federated learning, and proposes an evidence-based policy pathway tailored for developing countries, centered on synergistic governance, scalable infrastructure, and secure data ecosystems.
0 citationsRead paper