Selected work

Representative Papers

Trust Without Boundaries: An Architectural Analysis of Satellite Flight Software

Aug 14, 2026

This study addresses the insufficient security isolation caused by absent internal trust boundaries in satellite flight software. Focusing on NASA’s cFS architecture and leveraging NOS3 simulation alongside cross-framework comparisons, this work systematically exposes latent attack surfaces arising from shared authority within modular systems. Experimental results demonstrate that a single compromised component can masquerade as legitimate behavior to abuse privileges. Accordingly, this project proposes an architecture-level trust boundary reinforcement mechanism that effectively enhances onboard software security. By clarifying future improvement directions for flight software security architectures, this research fills a critical gap in systematic defense strategies for this domain.

0 citationsRead paper

Robustness of Presentation Attack Detection in Remote Identity Validation Scenarios

Jan 26, 2026

This study addresses the significant degradation in robustness of commercial presentation attack detection (PAD) systems for remote authentication under realistic environmental conditions, particularly low illumination and automated image capture. For the first time, it systematically quantifies the impact of these two common perturbations on the performance of mainstream PAD solutions by constructing ecologically valid test scenarios. Through error rate modeling and statistical analysis, the work evaluates shifts in classification accuracy across varying conditions. Results reveal that most systems exhibit approximately a fourfold increase in error rates under low-light conditions and a twofold increase under automated capture, with only one system maintaining a bona fide misclassification rate below 3% across all tested scenarios. The findings advocate for a new evaluation paradigm that mandates PAD robustness validation across diverse real-world settings.

0 citationsRead paper

DINO-Detect: A Simple yet Effective Framework for Blur-Robust AI-Generated Image Detection

Nov 16, 2025

Motion blur—arising from handheld capture or video compression—severely degrades the performance of existing AI-generated image (AIGI) detectors in real-world scenarios. To address this, we propose a teacher-student knowledge distillation framework specifically designed for motion-blur-robust AIGI detection. We freeze a high-capacity self-supervised teacher model (DINOv3) and leverage its rich semantic features and logit responses extracted from sharp images to supervise a lightweight student model trained directly on blurred images. Crucially, our method introduces dual-granularity distillation—jointly operating at both feature-level and logit-level—without requiring additional blur modeling or image enhancement. Evaluated on diverse synthetic multi-scale motion blur and realistic degraded datasets, our approach consistently outperforms state-of-the-art methods, achieving up to an 8.2% absolute improvement in detection accuracy. This demonstrates superior generalization across blur types and strong practical viability for real-world deployment.

0 citationsRead paper

Rethinking deep learning: linear regression remains a key benchmark in predicting terrestrial water storage

Oct 12, 2025

Deep learning models are often assumed superior to classical statistical methods in terrestrial water storage (TWS) forecasting, yet their practical advantages remain inadequately validated under complex hydrological regimes driven jointly by natural variability and anthropogenic interventions. Method: This study conducts a systematic, multi-scenario evaluation of LSTM and Temporal Fusion Transformer (TFT) against linear regression using the global HydroGlobe dataset, with rigorous out-of-sample validation across diverse hydroclimatic settings. Contribution/Results: Linear regression consistently outperforms both deep learning models across most metrics and scenarios, demonstrating superior robustness and generalizability. These findings challenge the prevailing assumption that deep learning inherently yields better hydrological forecasts, underscoring the necessity of establishing strong, interpretable baselines—particularly linear models—for fair model assessment. The study further advocates developing a standardized global TWS benchmark dataset explicitly encoding coupled natural–human drivers to enable scientifically grounded, reproducible model evaluation.

0 citationsRead paper

Recent publications

Latest Papers

Trust Without Boundaries: An Architectural Analysis of Satellite Flight Software

Aug 14, 2026

This study addresses the insufficient security isolation caused by absent internal trust boundaries in satellite flight software. Focusing on NASA’s cFS architecture and leveraging NOS3 simulation alongside cross-framework comparisons, this work systematically exposes latent attack surfaces arising from shared authority within modular systems. Experimental results demonstrate that a single compromised component can masquerade as legitimate behavior to abuse privileges. Accordingly, this project proposes an architecture-level trust boundary reinforcement mechanism that effectively enhances onboard software security. By clarifying future improvement directions for flight software security architectures, this research fills a critical gap in systematic defense strategies for this domain.

0 citationsRead paper

Robustness of Presentation Attack Detection in Remote Identity Validation Scenarios

Jan 26, 2026

This study addresses the significant degradation in robustness of commercial presentation attack detection (PAD) systems for remote authentication under realistic environmental conditions, particularly low illumination and automated image capture. For the first time, it systematically quantifies the impact of these two common perturbations on the performance of mainstream PAD solutions by constructing ecologically valid test scenarios. Through error rate modeling and statistical analysis, the work evaluates shifts in classification accuracy across varying conditions. Results reveal that most systems exhibit approximately a fourfold increase in error rates under low-light conditions and a twofold increase under automated capture, with only one system maintaining a bona fide misclassification rate below 3% across all tested scenarios. The findings advocate for a new evaluation paradigm that mandates PAD robustness validation across diverse real-world settings.

0 citationsRead paper

DINO-Detect: A Simple yet Effective Framework for Blur-Robust AI-Generated Image Detection

Nov 16, 2025

Motion blur—arising from handheld capture or video compression—severely degrades the performance of existing AI-generated image (AIGI) detectors in real-world scenarios. To address this, we propose a teacher-student knowledge distillation framework specifically designed for motion-blur-robust AIGI detection. We freeze a high-capacity self-supervised teacher model (DINOv3) and leverage its rich semantic features and logit responses extracted from sharp images to supervise a lightweight student model trained directly on blurred images. Crucially, our method introduces dual-granularity distillation—jointly operating at both feature-level and logit-level—without requiring additional blur modeling or image enhancement. Evaluated on diverse synthetic multi-scale motion blur and realistic degraded datasets, our approach consistently outperforms state-of-the-art methods, achieving up to an 8.2% absolute improvement in detection accuracy. This demonstrates superior generalization across blur types and strong practical viability for real-world deployment.

0 citationsRead paper

Rethinking deep learning: linear regression remains a key benchmark in predicting terrestrial water storage

Oct 12, 2025

Deep learning models are often assumed superior to classical statistical methods in terrestrial water storage (TWS) forecasting, yet their practical advantages remain inadequately validated under complex hydrological regimes driven jointly by natural variability and anthropogenic interventions. Method: This study conducts a systematic, multi-scenario evaluation of LSTM and Temporal Fusion Transformer (TFT) against linear regression using the global HydroGlobe dataset, with rigorous out-of-sample validation across diverse hydroclimatic settings. Contribution/Results: Linear regression consistently outperforms both deep learning models across most metrics and scenarios, demonstrating superior robustness and generalizability. These findings challenge the prevailing assumption that deep learning inherently yields better hydrological forecasts, underscoring the necessity of establishing strong, interpretable baselines—particularly linear models—for fair model assessment. The study further advocates developing a standardized global TWS benchmark dataset explicitly encoding coupled natural–human drivers to enable scientifically grounded, reproducible model evaluation.

0 citationsRead paper