TriSAR: Task Coordination and Collision Avoidance for Aerial Robot Teams in Disaster Response
研究通过比较遗传算法和贪婪分配策略在启用或禁用避碰情况下的表现，探讨了多无人机系统在灾害响应中的任务协调与碰撞避免问题。
0 citationsRead paper
Selected work
Representative Papers
From Thermal Preference Prediction to Adaptive Thermal Intervention: A Reinforcement Learning Approach Using Physiological and Environmental Sensing
本文通过结合多模态生理和环境传感与强化学习决策，提出了一种两阶段的个性化热舒适方法，以解决传统HVAC系统无法适应个体差异的问题。
0 citationsRead paper
An AI-Based Adaptive Learning Platform for Multilingual and Low-Resource Educational Contexts: A Case Study on Nigeria
This study addresses the scarcity of personalized learning and linguistic support in low-resource multilingual education by developing an AI-adaptive platform for Nigerian Pidgin. Through instruction tuning and multi-level quantization of large language models, we propose an experimentally validated adaptation framework for low-resource languages, introducing a novel dual-validation mechanism combining native-speaker cultural assessment with automated metrics. Our research elucidates the trade-offs between quantization bit-width, semantic quality, and inference latency, demonstrating that this approach significantly reduces computational overhead with minimal pedagogical degradation. Ultimately, this work enables the scalable deployment of intelligent educational systems that effectively balance cultural appropriateness with computational feasibility in resource-constrained settings.
0 citationsRead paper
Higher-order rich clubs and configuration models on general directed hypergraphs
本文提出了一种超富集俱乐部分析方法，通过超边编码高阶交互作用来检测复杂网络中的结构，适用于广义有向超图。
0 citationsRead paper
EmotionAI: A Privacy-Preserving Computational Intelligence Pipeline for Speech-Emotion-Grounded Conversational Analysis
This work addresses the privacy risks and manual effort inherent in traditional interview-based emotion analysis by proposing the first fully on-device, end-to-end emotion-driven conversational analysis framework. The system integrates speaker diarization, Whisper-based automatic speech recognition (ASR), and a wav2vec2-based emotion classifier to generate time-stamped emotional evidence—all processed locally without internet connectivity. A tri-model ensemble of local large language models then performs citation-constrained question-answering over this evidence. Evaluated on four subsets of RAVDESS, the approach achieves 48.8% emotion classification accuracy, significantly outperforming baseline methods. The entire pipeline runs on CPU with an average latency of 157 seconds (real-time factor 1.33), offering strong privacy guarantees, auditability, and cross-corpus emotional evidence integration, while candidly acknowledging its transfer limitations and the necessity of human validation boundaries.
0 citationsRead paper
Recent publications
Latest Papers
TriSAR: Task Coordination and Collision Avoidance for Aerial Robot Teams in Disaster Response
研究通过比较遗传算法和贪婪分配策略在启用或禁用避碰情况下的表现，探讨了多无人机系统在灾害响应中的任务协调与碰撞避免问题。
0 citationsRead paper
From Thermal Preference Prediction to Adaptive Thermal Intervention: A Reinforcement Learning Approach Using Physiological and Environmental Sensing
本文通过结合多模态生理和环境传感与强化学习决策，提出了一种两阶段的个性化热舒适方法，以解决传统HVAC系统无法适应个体差异的问题。
0 citationsRead paper
An AI-Based Adaptive Learning Platform for Multilingual and Low-Resource Educational Contexts: A Case Study on Nigeria
This study addresses the scarcity of personalized learning and linguistic support in low-resource multilingual education by developing an AI-adaptive platform for Nigerian Pidgin. Through instruction tuning and multi-level quantization of large language models, we propose an experimentally validated adaptation framework for low-resource languages, introducing a novel dual-validation mechanism combining native-speaker cultural assessment with automated metrics. Our research elucidates the trade-offs between quantization bit-width, semantic quality, and inference latency, demonstrating that this approach significantly reduces computational overhead with minimal pedagogical degradation. Ultimately, this work enables the scalable deployment of intelligent educational systems that effectively balance cultural appropriateness with computational feasibility in resource-constrained settings.
0 citationsRead paper
Higher-order rich clubs and configuration models on general directed hypergraphs
本文提出了一种超富集俱乐部分析方法，通过超边编码高阶交互作用来检测复杂网络中的结构，适用于广义有向超图。
0 citationsRead paper
EmotionAI: A Privacy-Preserving Computational Intelligence Pipeline for Speech-Emotion-Grounded Conversational Analysis
This work addresses the privacy risks and manual effort inherent in traditional interview-based emotion analysis by proposing the first fully on-device, end-to-end emotion-driven conversational analysis framework. The system integrates speaker diarization, Whisper-based automatic speech recognition (ASR), and a wav2vec2-based emotion classifier to generate time-stamped emotional evidence—all processed locally without internet connectivity. A tri-model ensemble of local large language models then performs citation-constrained question-answering over this evidence. Evaluated on four subsets of RAVDESS, the approach achieves 48.8% emotion classification accuracy, significantly outperforming baseline methods. The entire pipeline runs on CPU with an average latency of 157 seconds (real-time factor 1.33), offering strong privacy guarantees, auditability, and cross-corpus emotional evidence integration, while candidly acknowledging its transfer limitations and the necessity of human validation boundaries.
0 citationsRead paper