Selected work

Representative Papers

Balanced Soft mixture-of-expert model for Glaucoma Detection

Jul 28, 2026

This work addresses the challenges of imbalanced modality representations and insufficient joint optimization in multimodal glaucoma detection by proposing a Mixture-of-Experts (MoE) model integrating a soft gating mechanism with a load-balancing loss. The proposed approach effectively coordinates information fusion across three modalities, enabling the learning of more robust and discriminative joint representations. Evaluated on early glaucoma detection tasks, the method significantly outperforms existing unimodal, conventional multimodal, and state-of-the-art balanced multimodal approaches, achieving the highest reported AUC performance. Furthermore, the architecture demonstrates strong generalization capabilities and is readily adaptable to other ocular disease detection scenarios.

0 citationsRead paper

Leak-Free Cross-Validated Stacking with Per-Architecture Calibration for Sand-Boil Segmentation in Earthen Levees

Jul 28, 2026

This work addresses evaluation leakage in sand boil segmentation caused by data scarcity—such as synthetic images derived from the test set and overlap between ensemble weight tuning and test samples—by proposing a leakage-free cross-validated stacking framework. Within a five-fold cross-validation scheme, each fold rigorously excludes synthetic samples originating from the same source as the test images and aggregates predictions from five independently calibrated encoder-decoder models via a pixel-level meta-learner trained exclusively on out-of-fold data. Key innovations include a leakage-free evaluation protocol, an architecture-aware temperature calibration mechanism, and a mask-conditioned synthesis method requiring no additional annotations. Experiments reveal that high inter-model error correlation (0.894) limits stacking gains; nevertheless, the method achieves a mean IoU of 0.718 on an independent test set (averaged over three random seeds), substantially outperforming the original SandBoilNet (0.608).

0 citationsRead paper

Transition-Aware Backend Dispatch for Edge LLM Inference

Jul 19, 2026

This work addresses the challenges of deploying large language models on edge devices, where inference performance is constrained by model size and significant performance variations across backends due to tensor shape differences. Static scheduling fails to adapt to runtime dynamics, while per-operator backend selection incurs prohibitive switching overhead. To overcome these limitations, the authors propose a dynamic backend scheduling strategy that jointly models operator shape characteristics and cross-backend switching context for the first time, thereby preserving shape-aware performance gains while minimizing switching costs. Evaluation across multiple backends—PyTorch (CPU/CUDA) and ONNX Runtime (CPU)—demonstrates that, over 9,584 operator instances, the approach reduces average latency by 17.4%, energy consumption by 14.4%, and energy-delay product by 28.5% compared to the best static strategy, with consistent generalization improvements observed across six to seven Transformer-based models.

0 citationsRead paper

MTD-Playground: An Attacker-Aware Evaluation Framework for Network Moving Target Defense

Jul 13, 2026

This study addresses the fragmentation and limited reproducibility in existing Moving Target Defense (MTD) evaluations, which stem from the absence of a unified attacker model, consistent scenarios, and standardized metrics. To overcome these limitations, this work proposes MTD-Playground—the first comprehensive, attacker-oriented evaluation framework designed for benchmarking SDN-based path randomization MTD techniques within enterprise-level multi-stage attack scenarios. The framework introduces an innovative composite evaluation paradigm that jointly considers security effectiveness, system performance, and deployment feasibility. Experimental results demonstrate that aggressive mutation strategies can reduce attack success rates to 4–20%, extend attack completion times to 160–311 seconds, increase network throughput by 30.9%, decrease internal path latency, and achieve these gains without causing service disruption.

0 citationsRead paper

Multi-Conditioned Diffusion Synthesis of Sand Boils for Low-Resource Earthen-Levee Inspection

Jul 07, 2026

This work addresses the scarcity of real annotated data for detecting sand boil defects in earthen levees by proposing a diffusion model–based synthetic data augmentation approach. Leveraging Stable Diffusion XL fine-tuned with DreamBooth, the method integrates multi-branch ControlNet, Prompt Atlas, and soft-mask inpainting to repaint backgrounds while preserving authentic defect structures, thereby avoiding stitching artifacts and enabling cross-category transfer. Requiring only a limited number of real samples, the framework generates high-quality, label-traceable synthetic images. From an initial set of 1,020 candidates, 815 high-fidelity samples were selected, achieving a strong balance among fidelity, diversity, and label reliability. The study also releases preset configurations and a hybrid augmented dataset to support future research.

0 citationsRead paper

Recent publications

Latest Papers

Balanced Soft mixture-of-expert model for Glaucoma Detection

Jul 28, 2026

This work addresses the challenges of imbalanced modality representations and insufficient joint optimization in multimodal glaucoma detection by proposing a Mixture-of-Experts (MoE) model integrating a soft gating mechanism with a load-balancing loss. The proposed approach effectively coordinates information fusion across three modalities, enabling the learning of more robust and discriminative joint representations. Evaluated on early glaucoma detection tasks, the method significantly outperforms existing unimodal, conventional multimodal, and state-of-the-art balanced multimodal approaches, achieving the highest reported AUC performance. Furthermore, the architecture demonstrates strong generalization capabilities and is readily adaptable to other ocular disease detection scenarios.

0 citationsRead paper

Leak-Free Cross-Validated Stacking with Per-Architecture Calibration for Sand-Boil Segmentation in Earthen Levees

Jul 28, 2026

This work addresses evaluation leakage in sand boil segmentation caused by data scarcity—such as synthetic images derived from the test set and overlap between ensemble weight tuning and test samples—by proposing a leakage-free cross-validated stacking framework. Within a five-fold cross-validation scheme, each fold rigorously excludes synthetic samples originating from the same source as the test images and aggregates predictions from five independently calibrated encoder-decoder models via a pixel-level meta-learner trained exclusively on out-of-fold data. Key innovations include a leakage-free evaluation protocol, an architecture-aware temperature calibration mechanism, and a mask-conditioned synthesis method requiring no additional annotations. Experiments reveal that high inter-model error correlation (0.894) limits stacking gains; nevertheless, the method achieves a mean IoU of 0.718 on an independent test set (averaged over three random seeds), substantially outperforming the original SandBoilNet (0.608).

0 citationsRead paper

Transition-Aware Backend Dispatch for Edge LLM Inference

Jul 19, 2026

This work addresses the challenges of deploying large language models on edge devices, where inference performance is constrained by model size and significant performance variations across backends due to tensor shape differences. Static scheduling fails to adapt to runtime dynamics, while per-operator backend selection incurs prohibitive switching overhead. To overcome these limitations, the authors propose a dynamic backend scheduling strategy that jointly models operator shape characteristics and cross-backend switching context for the first time, thereby preserving shape-aware performance gains while minimizing switching costs. Evaluation across multiple backends—PyTorch (CPU/CUDA) and ONNX Runtime (CPU)—demonstrates that, over 9,584 operator instances, the approach reduces average latency by 17.4%, energy consumption by 14.4%, and energy-delay product by 28.5% compared to the best static strategy, with consistent generalization improvements observed across six to seven Transformer-based models.

0 citationsRead paper

MTD-Playground: An Attacker-Aware Evaluation Framework for Network Moving Target Defense

Jul 13, 2026

This study addresses the fragmentation and limited reproducibility in existing Moving Target Defense (MTD) evaluations, which stem from the absence of a unified attacker model, consistent scenarios, and standardized metrics. To overcome these limitations, this work proposes MTD-Playground—the first comprehensive, attacker-oriented evaluation framework designed for benchmarking SDN-based path randomization MTD techniques within enterprise-level multi-stage attack scenarios. The framework introduces an innovative composite evaluation paradigm that jointly considers security effectiveness, system performance, and deployment feasibility. Experimental results demonstrate that aggressive mutation strategies can reduce attack success rates to 4–20%, extend attack completion times to 160–311 seconds, increase network throughput by 30.9%, decrease internal path latency, and achieve these gains without causing service disruption.

0 citationsRead paper

Multi-Conditioned Diffusion Synthesis of Sand Boils for Low-Resource Earthen-Levee Inspection

Jul 07, 2026

This work addresses the scarcity of real annotated data for detecting sand boil defects in earthen levees by proposing a diffusion model–based synthetic data augmentation approach. Leveraging Stable Diffusion XL fine-tuned with DreamBooth, the method integrates multi-branch ControlNet, Prompt Atlas, and soft-mask inpainting to repaint backgrounds while preserving authentic defect structures, thereby avoiding stitching artifacts and enabling cross-category transfer. Requiring only a limited number of real samples, the framework generates high-quality, label-traceable synthetic images. From an initial set of 1,020 candidates, 815 high-fidelity samples were selected, achieving a strong balance among fidelity, diversity, and label reliability. The study also releases preset configurations and a hybrid augmented dataset to support future research.

0 citationsRead paper