Institution profile

Wuhan University

Academic institutionasia · cn
Official website
Research library1,741linked papers
Opportunities0open roles
Selected work

Representative Papers

MambaHSI: Spatial–Spectral Mamba for Hyperspectral Image Classification

Jan 09, 2025IEEE Transactions on Geoscience and Remote Sensing

To address the high computational cost of Transformers and the difficulty of existing Mamba architectures in jointly modeling spatial and spectral structures in hyperspectral image (HSI) classification, this paper proposes HS-Mamba—the first full-image-level Mamba architecture for HSI. Methodologically, it introduces (1) decoupled Spatial Mamba Blocks (SpaMB) and Spectral Mamba Blocks (SpeMB) to capture long-range spatial dependencies and grouped spectral correlations, respectively; (2) a Spatial-Spectral Fusion Module (SSFM) for adaptive cross-domain feature interaction; and (3) linear-complexity state-space modeling throughout the network to balance efficiency and representational capacity. Evaluated on four benchmark HSI datasets—including Indian Pines—HS-Mamba achieves state-of-the-art accuracy with significantly fewer parameters and lower FLOPs than CNNs, Transformers, and prior Mamba variants, demonstrating Mamba’s feasibility and superiority as an efficient backbone for HSI classification.

15 citationsRead paper

Rethinking Membership Inference Attacks Against Transfer Learning

Jan 20, 2025IEEE Transactions on Information Forensics and Security

In transfer learning, teacher models’ training data are vulnerable to membership inference attacks (MIAs), yet existing work predominantly assumes attackers have black-box or white-box access to the teacher model—overlooking privacy threats when only the student model is accessible in a white-box setting. Method: This paper identifies that representational discrepancies between teacher and student hidden layers can be exploited for MIAs, and proposes the first MIA framework requiring only white-box access to the student model. It employs shadow models to calibrate student-layer representations, models teacher–student representation divergence, and infers teacher training membership via reverse inference—without requiring the teacher model itself. Contribution/Results: The method achieves high inference accuracy across four benchmark datasets and diverse transfer learning tasks. It significantly expands the threat surface of MIAs in transfer learning and provides novel empirical evidence and insights for safeguarding teacher model privacy.

13 citationsRead paper

Understanding the Issues, Their Causes and Solutions in Microservices Systems: An Empirical Study

Feb 03, 2023arXiv.org

Microservice system developers lack empirical evidence regarding the types, root causes, and remediation strategies of recurring issues. Method: We adopt a mixed-methods approach—quantitatively analyzing 2,641 open-source issues, qualitatively interviewing 15 practitioners, and conducting a global survey with 150 practitioners. Contribution/Results: We introduce the first comprehensive, domain-specific three-level taxonomy (“Issue–Cause–Solution”) for microservices. We identify five high-frequency issue domains—including technical debt, CI/CD pipeline failures, and exception handling—and three predominant root causes, notably generic programming errors. From our analysis, we distill 177 actionable, context-aware remediation strategies. This work establishes an empirical foundation for microservice fault diagnosis and mitigation, delivers practical guidance for industry practitioners, and pinpoints critical research directions for next-generation microservice engineering.

6 citations2 influentialRead paper

Controlled LLM Training on Spectral Sphere

Jan 13, 2026

Existing large-model optimizers struggle to simultaneously stabilize both weights and their updates, often leading to issues such as activation blowup, slow convergence, and imbalanced expert utilization in Mixture-of-Experts (MoE) architectures. This work proposes the Spectral Sphere Optimizer (SSO), which introduces, for the first time, module-wise joint spectral constraints on both weights and their updates, rigorously aligning with Maximal Update Parametrization (μP) to ensure optimization stability. SSO is derived from the steepest descent direction on the spectral sphere and integrates seamlessly into the Megatron framework, supporting Dense, MoE, and DeepNet architectures. Experiments demonstrate that SSO consistently outperforms AdamW and Muon across a 1.7B Dense model, an 8B-A1B MoE model, and a 200-layer DeepNet, effectively suppressing anomalous activations, improving routing balance, and enhancing training stability and scalability.

5 citationsRead paper

Enhancing Adversarial Transferability Through Neighborhood Conditional Sampling

May 25, 2024arXiv.org

Existing flatness-enhancement methods suffer from inconsistent flatness definitions, heuristic attack designs, and a lack of theoretical grounding and optimization guarantees, thereby limiting the efficacy and efficiency of black-box transfer attacks. To address this, we propose the Neighborhood-Conditional Sampling (NCS) framework: (i) it introduces a novel max-min bilevel optimization to explicitly maximize flatness; (ii) it designs a zero-overhead momentum-based prior-gradient inversion approximation (PGIA); and (iii) it unifies multiple state-of-the-art attacks as special cases. We theoretically establish that NCS’s flatness regularization and neighborhood sampling operate synergistically. Empirically, NCS achieves significantly higher transfer success rates while maintaining lightweight computation—only 50% of the cost of current SOTA methods. As a plug-and-play module, NCS is universally applicable for enhancing diverse adversarial attacks.

4 citations1 influentialRead paper
Recent publications

Latest Papers