Context operations to architecture modelling output from large language models and evaluation criteria for their use in systems engineering design
本文提出一种框架，通过模块化上下文单元来组装大型语言模型的工程设计输出，并提供了一种评估方法以确保其符合设计意图。
0 citationsRead paper
Selected work
Representative Papers
Self Supervised Learning from Automatically Generated Demonstrations for Visual Robotic Manipulation
This work proposes a self-supervised visual manipulation method that eliminates the need for manual programming, human demonstrations, or extrinsic camera-robot calibration, which traditionally incur high deployment costs. The robot autonomously generates visual demonstrations near the target pose and learns relative pose corrections directly from wrist-mounted RGB images. A coarse-to-fine two-stage control strategy is employed, and an image-pose paired dataset is constructed using ROS 2 and Isaac Sim. A convolutional neural network regresses relative translation and rotation from single-frame RGB inputs. In simulation, the planar positioning error decreases from 9.69 mm to 5.38 mm. On a real UR5e robot, the method achieves grasping success rates of 66.6% and 63.6% on two object categories and demonstrates robustness under rotational perturbations.
0 citationsRead paper
Unit-Modified Weibull Distribution and Quantile Regression Model
Many Sustainable Development Goals (SDGs) indicators—such as health coverage and education completion rates—are bounded in the unit interval (0,1), yet lack appropriate probabilistic models tailored to such constrained continuous outcomes. Method: We propose the Unit-Modified Weibull (UMW) distribution—the first unit-interval distribution derived from the modified Weibull via probability integral transformation—and develop a quantile regression framework parameterized by quantiles, enabling flexible modeling of doubly bounded data. Estimation employs maximum likelihood, with small-sample robustness validated via Monte Carlo simulation. Results: Empirical applications to SDG 3 and SDG 4 indicators, as well as literacy data from children with dyslexia, demonstrate superior goodness-of-fit and interpretability. The UMW quantile regression significantly extends the statistical toolkit for analyzing bounded continuous outcomes in sustainability science.
0 citationsRead paper
On the Performance of Cyber-Biomedical Features for Intrusion Detection in Healthcare 5.0
Healthcare 5.0’s hyperconnectivity intensifies cybersecurity threats, yet existing AI-based intrusion detection models largely neglect biomedical data, compromising detection efficacy and interpretability. To address this gap, we propose a novel multi-source heterogeneous modeling framework that jointly leverages network traffic and biomedical sensor data (e.g., body temperature). Our method employs XGBoost for classification and SHAP for post-hoc interpretability analysis. We are the first to systematically demonstrate the discriminative value of physiological signals—particularly temperature—in detecting medical cyberattacks. Results show that temperature exhibits high explanatory power for spoofing attacks (Shapley value = 0.37). On a hybrid dataset, the model achieves an F1-score of 99% for benign and data-tampering instances, and 81% for spoofing attacks. Network features provide robust general detection capability, while physiological features significantly enhance both interpretability and robustness for attack-specific identification.
0 citationsRead paper
MixFunn: A Neural Network for Differential Equations with Improved Generalization and Interpretability
Existing differential equation solvers face limitations in accuracy, generalization, and interpretability. To address these challenges, this paper proposes MixFunn—a novel neural network architecture featuring hybrid functional neurons and second-order neurons, integrated within the physics-informed neural networks (PINNs) framework. This design markedly enhances representational capacity while reducing parameter count by four orders of magnitude. Crucially, MixFunn enables direct extraction of closed-form analytical approximations from the trained model. Evaluated across diverse partial differential equations from classical mechanics, quantum mechanics, and fluid dynamics, MixFunn achieves significantly higher solution accuracy and demonstrates strong out-of-distribution generalization. The core contribution lies in unifying high accuracy, lightweight modeling, and inherent interpretability—establishing a new paradigm for scientific machine learning that bridges theoretical rigor and engineering practicality.
0 citationsRead paper
Recent publications
Latest Papers
Context operations to architecture modelling output from large language models and evaluation criteria for their use in systems engineering design
本文提出一种框架，通过模块化上下文单元来组装大型语言模型的工程设计输出，并提供了一种评估方法以确保其符合设计意图。
0 citationsRead paper
Self Supervised Learning from Automatically Generated Demonstrations for Visual Robotic Manipulation
This work proposes a self-supervised visual manipulation method that eliminates the need for manual programming, human demonstrations, or extrinsic camera-robot calibration, which traditionally incur high deployment costs. The robot autonomously generates visual demonstrations near the target pose and learns relative pose corrections directly from wrist-mounted RGB images. A coarse-to-fine two-stage control strategy is employed, and an image-pose paired dataset is constructed using ROS 2 and Isaac Sim. A convolutional neural network regresses relative translation and rotation from single-frame RGB inputs. In simulation, the planar positioning error decreases from 9.69 mm to 5.38 mm. On a real UR5e robot, the method achieves grasping success rates of 66.6% and 63.6% on two object categories and demonstrates robustness under rotational perturbations.
0 citationsRead paper
Unit-Modified Weibull Distribution and Quantile Regression Model
Many Sustainable Development Goals (SDGs) indicators—such as health coverage and education completion rates—are bounded in the unit interval (0,1), yet lack appropriate probabilistic models tailored to such constrained continuous outcomes. Method: We propose the Unit-Modified Weibull (UMW) distribution—the first unit-interval distribution derived from the modified Weibull via probability integral transformation—and develop a quantile regression framework parameterized by quantiles, enabling flexible modeling of doubly bounded data. Estimation employs maximum likelihood, with small-sample robustness validated via Monte Carlo simulation. Results: Empirical applications to SDG 3 and SDG 4 indicators, as well as literacy data from children with dyslexia, demonstrate superior goodness-of-fit and interpretability. The UMW quantile regression significantly extends the statistical toolkit for analyzing bounded continuous outcomes in sustainability science.
0 citationsRead paper
On the Performance of Cyber-Biomedical Features for Intrusion Detection in Healthcare 5.0
Healthcare 5.0’s hyperconnectivity intensifies cybersecurity threats, yet existing AI-based intrusion detection models largely neglect biomedical data, compromising detection efficacy and interpretability. To address this gap, we propose a novel multi-source heterogeneous modeling framework that jointly leverages network traffic and biomedical sensor data (e.g., body temperature). Our method employs XGBoost for classification and SHAP for post-hoc interpretability analysis. We are the first to systematically demonstrate the discriminative value of physiological signals—particularly temperature—in detecting medical cyberattacks. Results show that temperature exhibits high explanatory power for spoofing attacks (Shapley value = 0.37). On a hybrid dataset, the model achieves an F1-score of 99% for benign and data-tampering instances, and 81% for spoofing attacks. Network features provide robust general detection capability, while physiological features significantly enhance both interpretability and robustness for attack-specific identification.
0 citationsRead paper
MixFunn: A Neural Network for Differential Equations with Improved Generalization and Interpretability
Existing differential equation solvers face limitations in accuracy, generalization, and interpretability. To address these challenges, this paper proposes MixFunn—a novel neural network architecture featuring hybrid functional neurons and second-order neurons, integrated within the physics-informed neural networks (PINNs) framework. This design markedly enhances representational capacity while reducing parameter count by four orders of magnitude. Crucially, MixFunn enables direct extraction of closed-form analytical approximations from the trained model. Evaluated across diverse partial differential equations from classical mechanics, quantum mechanics, and fluid dynamics, MixFunn achieves significantly higher solution accuracy and demonstrates strong out-of-distribution generalization. The core contribution lies in unifying high accuracy, lightweight modeling, and inherent interpretability—establishing a new paradigm for scientific machine learning that bridges theoretical rigor and engineering practicality.
0 citationsRead paper