Selected work

Representative Papers

A Disturbance in the Force: Force Actuation on the RAVEN II Surgical Robot with Parallel Motor-Cable Units

Aug 06, 2026

This study addresses the longstanding lack of effective haptic feedback in surgical robots and the challenge of acquiring high-quality training data under representative external forces without adding sensors. To this end, the authors propose a non-invasive, six-degree-of-freedom parallel motor-cable system arranged around the workspace of the RAVEN II surgical robot. By applying precisely controlled cable tensions to the robot’s end-effector, the system delivers accurate external forces without impeding its motion. The framework integrates custom motor hardware, a tension control algorithm, sensor drivers, and a simulation module, enabling—for the first time—high-precision force application using a parallel cable-driven architecture. Experimental results demonstrate a force control error of less than 1 N, providing high-fidelity training data for force-perception learning in robotic surgery.

0 citationsRead paper

Predicting COVID-19 Prevalence Using Wastewater RNA Surveillance: A Semi-Supervised Learning Approach with Temporal Feature Trust

Nov 27, 2025

Following COVID-19’s transition to endemicity, wastewater RNA surveillance data exhibit heterogeneous temporal reliability due to variations in sampling protocols, assay sensitivity, and evolving transmission dynamics. Method: We propose a semi-supervised deep neural network that integrates temporal feature credibility modeling. It uses wastewater viral RNA concentrations as primary input—augmented with confounding factors—to learn a nonlinear mapping for daily case forecasting. Crucially, it introduces a dynamic feature reliability weighting mechanism that explicitly quantifies sample quality across epidemic phases and employs semi-supervised learning to mitigate scarcity of high-confidence labeled data. Results: Experiments demonstrate significantly improved generalization under data quality fluctuations: the model achieves high-accuracy daily predictions (MAE < 8.2) during high-reliability periods. It establishes a novel, interpretable, and robust paradigm for non-invasive epidemiological trend monitoring.

0 citationsRead paper

What Does Normal Even Mean? Evaluating Benign Traffic in Intrusion Detection Datasets

Sep 11, 2025

Conventional intrusion detection systems label all benign traffic as a single class, introducing semantic ambiguity that obscures inherent structural heterogeneity. Method: To investigate whether benign traffic exhibits distinguishable fine-grained substructure, we apply unsupervised clustering algorithms—including HDBSCAN and Mean Shift—to benign samples from NSL-KDD, UNSW-NB15, and CIC-IDS2017. Contribution/Results: Empirical analysis reveals statistically significant and stable clustering patterns across all datasets, confirming intrinsic semantic heterogeneity within benign traffic. We thus propose a novel paradigm—“fine-grained partitioning of benign traffic”—which challenges the long-standing single-class labeling assumption. Subsequent experiments demonstrate that classifiers trained with multi-subclass benign labels—derived from the discovered structure—achieve improved attack-type discrimination and enhanced overall multiclass classification performance, particularly in distinguishing subtle or zero-day attacks. This work establishes a foundation for semantically aware benign modeling in intrusion detection.

0 citationsRead paper

Recent publications

Latest Papers

A Disturbance in the Force: Force Actuation on the RAVEN II Surgical Robot with Parallel Motor-Cable Units

Aug 06, 2026

This study addresses the longstanding lack of effective haptic feedback in surgical robots and the challenge of acquiring high-quality training data under representative external forces without adding sensors. To this end, the authors propose a non-invasive, six-degree-of-freedom parallel motor-cable system arranged around the workspace of the RAVEN II surgical robot. By applying precisely controlled cable tensions to the robot’s end-effector, the system delivers accurate external forces without impeding its motion. The framework integrates custom motor hardware, a tension control algorithm, sensor drivers, and a simulation module, enabling—for the first time—high-precision force application using a parallel cable-driven architecture. Experimental results demonstrate a force control error of less than 1 N, providing high-fidelity training data for force-perception learning in robotic surgery.

0 citationsRead paper

Predicting COVID-19 Prevalence Using Wastewater RNA Surveillance: A Semi-Supervised Learning Approach with Temporal Feature Trust

Nov 27, 2025

Following COVID-19’s transition to endemicity, wastewater RNA surveillance data exhibit heterogeneous temporal reliability due to variations in sampling protocols, assay sensitivity, and evolving transmission dynamics. Method: We propose a semi-supervised deep neural network that integrates temporal feature credibility modeling. It uses wastewater viral RNA concentrations as primary input—augmented with confounding factors—to learn a nonlinear mapping for daily case forecasting. Crucially, it introduces a dynamic feature reliability weighting mechanism that explicitly quantifies sample quality across epidemic phases and employs semi-supervised learning to mitigate scarcity of high-confidence labeled data. Results: Experiments demonstrate significantly improved generalization under data quality fluctuations: the model achieves high-accuracy daily predictions (MAE < 8.2) during high-reliability periods. It establishes a novel, interpretable, and robust paradigm for non-invasive epidemiological trend monitoring.

0 citationsRead paper

What Does Normal Even Mean? Evaluating Benign Traffic in Intrusion Detection Datasets

Sep 11, 2025

Conventional intrusion detection systems label all benign traffic as a single class, introducing semantic ambiguity that obscures inherent structural heterogeneity. Method: To investigate whether benign traffic exhibits distinguishable fine-grained substructure, we apply unsupervised clustering algorithms—including HDBSCAN and Mean Shift—to benign samples from NSL-KDD, UNSW-NB15, and CIC-IDS2017. Contribution/Results: Empirical analysis reveals statistically significant and stable clustering patterns across all datasets, confirming intrinsic semantic heterogeneity within benign traffic. We thus propose a novel paradigm—“fine-grained partitioning of benign traffic”—which challenges the long-standing single-class labeling assumption. Subsequent experiments demonstrate that classifiers trained with multi-subclass benign labels—derived from the discovered structure—achieve improved attack-type discrimination and enhanced overall multiclass classification performance, particularly in distinguishing subtle or zero-day attacks. This work establishes a foundation for semantically aware benign modeling in intrusion detection.

0 citationsRead paper