The widening evaluation gap in medical large language model research 2023 to 2026
研究探讨了2023至2026年间医学大语言模型评价滞后的现象，通过分析PubMed数据指出随机对照试验等设计的滞后加剧，并提出模型更新速度与临床证据生成速度不匹配的问题。
0 citationsRead paper
Selected work
Representative Papers
Target leakage, not model class, explains reported accuracy in survey-based cardiovascular screening: a leakage-tiered audit of glass-box and tabular foundation models
研究通过分层审计方法，揭示心血管筛查模型的高准确性主要源于目标泄漏而非模型本身，并评估了透明模型在公平性和不确定性处理上的优势。
0 citationsRead paper
Bulbul: A Dataset for Dialectal Arabic Speech Recognition
为解决阿拉伯语自动语音识别面临的独特挑战，通过构建包含275名来自11个阿拉伯国家的说话者的多方言数据集BULBUL，并采用两层人工验证确保质量。
0 citationsRead paper
Expertise-Based Developer Assignment for Long-Term Software Components in Open-Source Projects
This study addresses the lack of effective mechanisms for matching developer expertise to new long-term software components in open-source projects, a gap that leads to inefficient task assignment and degraded code maintainability. Focusing specifically on the allocation of developers to long-term modules, this work proposes a novel approach that constructs a developer expertise model based on historical Git commits and integrates project structure to enable precise recommendations. The method combines commit data analysis, expertise modeling, model inference, and server-side caching. Experimental results demonstrate that 72.4% of target developers were ranked within the top 10 recommendations out of a pool of 47 candidates, and the incorporation of caching improved worst-case response time by a factor of 9.86.
0 citationsRead paper
Optimizing IoT Intrusion Detection with Tabular Foundation Models for Smart City Forensics
This study addresses the dual demands of real-time performance and high accuracy in intrusion detection for smart city IoT environments, where conventional high-accuracy ensemble methods like Random Forest are often impractical due to excessive computational overhead. To bridge this gap, the work introduces TabPFNv2.5—a tabular foundation model—into IoT security forensics for the first time, proposing a hybrid detection architecture that combines rapid initial screening with refined classification. Specifically, TabPFNv2.5 performs efficient preliminary filtering, followed by an ensemble model for precise final decisions. Experiments on the TON IoT dataset demonstrate that TabPFNv2.5 achieves 40× faster inference than Random Forest while maintaining a binary classification accuracy of 97%. The study also identifies a performance bottleneck in scan attack detection (F1 = 69.8%) and highlights the critical role of feature similarity in enabling cross-device generalization.
0 citationsRead paper
Recent publications
Latest Papers
The widening evaluation gap in medical large language model research 2023 to 2026
研究探讨了2023至2026年间医学大语言模型评价滞后的现象，通过分析PubMed数据指出随机对照试验等设计的滞后加剧，并提出模型更新速度与临床证据生成速度不匹配的问题。
0 citationsRead paper
Target leakage, not model class, explains reported accuracy in survey-based cardiovascular screening: a leakage-tiered audit of glass-box and tabular foundation models
研究通过分层审计方法，揭示心血管筛查模型的高准确性主要源于目标泄漏而非模型本身，并评估了透明模型在公平性和不确定性处理上的优势。
0 citationsRead paper
Bulbul: A Dataset for Dialectal Arabic Speech Recognition
为解决阿拉伯语自动语音识别面临的独特挑战，通过构建包含275名来自11个阿拉伯国家的说话者的多方言数据集BULBUL，并采用两层人工验证确保质量。
0 citationsRead paper
Expertise-Based Developer Assignment for Long-Term Software Components in Open-Source Projects
This study addresses the lack of effective mechanisms for matching developer expertise to new long-term software components in open-source projects, a gap that leads to inefficient task assignment and degraded code maintainability. Focusing specifically on the allocation of developers to long-term modules, this work proposes a novel approach that constructs a developer expertise model based on historical Git commits and integrates project structure to enable precise recommendations. The method combines commit data analysis, expertise modeling, model inference, and server-side caching. Experimental results demonstrate that 72.4% of target developers were ranked within the top 10 recommendations out of a pool of 47 candidates, and the incorporation of caching improved worst-case response time by a factor of 9.86.
0 citationsRead paper
Optimizing IoT Intrusion Detection with Tabular Foundation Models for Smart City Forensics
This study addresses the dual demands of real-time performance and high accuracy in intrusion detection for smart city IoT environments, where conventional high-accuracy ensemble methods like Random Forest are often impractical due to excessive computational overhead. To bridge this gap, the work introduces TabPFNv2.5—a tabular foundation model—into IoT security forensics for the first time, proposing a hybrid detection architecture that combines rapid initial screening with refined classification. Specifically, TabPFNv2.5 performs efficient preliminary filtering, followed by an ensemble model for precise final decisions. Experiments on the TON IoT dataset demonstrate that TabPFNv2.5 achieves 40× faster inference than Random Forest while maintaining a binary classification accuracy of 97%. The study also identifies a performance bottleneck in scan attack detection (F1 = 69.8%) and highlights the critical role of feature similarity in enabling cross-device generalization.
0 citationsRead paper