Selected work

Representative Papers

Protecting patient privacy in clinical foundation models: Technical and legal perspectives

Aug 07, 2026

This work addresses the underexplored risk of indirect patient privacy leakage through deployed clinical foundation models, a challenge inadequately mitigated by current legal and technical safeguards. It proposes, for the first time, a context-aware privacy risk assessment framework that systematically integrates technical and legal perspectives to evaluate indirect leakage risks across the model’s entire lifecycle. By combining privacy risk modeling, leakage scenario simulation, and compliance mapping with technical measures such as differential privacy and access control, the framework elucidates representative leakage mechanisms. It delivers an actionable risk assessment workflow and cross-jurisdictional compliance guidance, enabling robust privacy protection without compromising model utility.

0 citationsRead paper

Private Vertical Federated Inference for Time-Series

May 08, 2026

This work addresses the challenges of high computational overhead and embedding leakage in privacy-sensitive vertical federated learning (VFL) for sequential data inference. To this end, the authors propose PPHH-VFL, a hybrid architecture that splits the model head into a plaintext public head and a lightweight private head secured via secure multi-party computation (MPC), while incorporating adversarial training to protect the privacy of public embeddings. This approach is the first to simultaneously achieve high efficiency and strong privacy guarantees in vertical federated sequential inference. Experimental results demonstrate that PPHH-VFL accelerates inference by up to one million times compared to end-to-end MPC, and achieves a 44.4× speedup over a VFL+MPC baseline under wide-area network conditions, reducing communication overhead by 91.2% (from 1.7 GB to 19 MB per batch), while improving classification accuracy by 2.50% and lowering regression RMSE by 40.7%.

0 citationsRead paper

Can LLMs Take Retrieved Information with a Grain of Salt?

May 07, 2026

This work addresses the critical challenge that large language models (LLMs) often fail to modulate the certainty of their responses in accordance with the uncertainty inherent in retrieved contextual information—a shortcoming that poses significant risks in high-stakes domains such as healthcare and finance. To this end, the authors introduce the first evaluation metric specifically designed to assess contextual certainty adherence, systematically exposing deficiencies in current LLMs on this task. They further propose a general-purpose prompting strategy that requires no model weight modifications, integrating prior-knowledge reminders, certainty calibration, and context simplification mechanisms. Experimental results demonstrate that this approach reduces certainty adherence errors by 25% on average across multiple mainstream LLMs, substantially enhancing their ability to respond cautiously when confronted with uncertain input contexts.

0 citationsRead paper

LassoFlexNet: Flexible Neural Architecture for Tabular Data

Mar 20, 2026

Deep neural networks often underperform tree-based models on tabular data due to irrelevant features, feature heterogeneity, and local irregularities. To address this, this work proposes the LassoFlexNet architecture, which enhances model expressivity through five inductive biases: it employs per-feature embeddings to separately capture linear and nonlinear marginal contributions, integrates a grouped Lasso penalty for sparse variable selection, and introduces a novel Sequential Hierarchical Proximal Adaptive Gradient optimizer incorporating exponential moving averages (EMA). This design effectively breaks undesirable rotational invariance inherent in standard neural networks. Evaluated across 52 benchmark datasets, LassoFlexNet matches or surpasses leading tree-based models, achieving up to a 10% relative performance gain while maintaining strong interpretability.

0 citationsRead paper

On the Trade-Off Between Transparency and Security in Adversarial Machine Learning

Nov 14, 2025

This work investigates the fundamental tension between model transparency and system security in adversarial machine learning. Focusing on transferable adversarial example attacks, we formulate a game-theoretic framework—incorporating both Nash and Stackelberg equilibria—between attackers and defenders, and conduct large-scale empirical evaluation across nine attack methods and 181 models. Our analysis reveals that transparency substantially increases attack success: merely knowing whether a model is protected suffices to significantly degrade its robustness. We introduce “stealthiness” as a critical defensive advantage, formally demonstrating that strategically concealing defense mechanisms yields superior security outcomes compared to full transparency. To our knowledge, this is the first study to formalize the transparency–security trade-off using rigorous game-theoretic analysis. The findings provide both theoretical grounding and practical guidance for deploying responsible AI systems under adversarial conditions.

0 citationsRead paper

Recent publications

Latest Papers

Protecting patient privacy in clinical foundation models: Technical and legal perspectives

Aug 07, 2026

This work addresses the underexplored risk of indirect patient privacy leakage through deployed clinical foundation models, a challenge inadequately mitigated by current legal and technical safeguards. It proposes, for the first time, a context-aware privacy risk assessment framework that systematically integrates technical and legal perspectives to evaluate indirect leakage risks across the model’s entire lifecycle. By combining privacy risk modeling, leakage scenario simulation, and compliance mapping with technical measures such as differential privacy and access control, the framework elucidates representative leakage mechanisms. It delivers an actionable risk assessment workflow and cross-jurisdictional compliance guidance, enabling robust privacy protection without compromising model utility.

0 citationsRead paper

Private Vertical Federated Inference for Time-Series

May 08, 2026

This work addresses the challenges of high computational overhead and embedding leakage in privacy-sensitive vertical federated learning (VFL) for sequential data inference. To this end, the authors propose PPHH-VFL, a hybrid architecture that splits the model head into a plaintext public head and a lightweight private head secured via secure multi-party computation (MPC), while incorporating adversarial training to protect the privacy of public embeddings. This approach is the first to simultaneously achieve high efficiency and strong privacy guarantees in vertical federated sequential inference. Experimental results demonstrate that PPHH-VFL accelerates inference by up to one million times compared to end-to-end MPC, and achieves a 44.4× speedup over a VFL+MPC baseline under wide-area network conditions, reducing communication overhead by 91.2% (from 1.7 GB to 19 MB per batch), while improving classification accuracy by 2.50% and lowering regression RMSE by 40.7%.

0 citationsRead paper

Can LLMs Take Retrieved Information with a Grain of Salt?

May 07, 2026

This work addresses the critical challenge that large language models (LLMs) often fail to modulate the certainty of their responses in accordance with the uncertainty inherent in retrieved contextual information—a shortcoming that poses significant risks in high-stakes domains such as healthcare and finance. To this end, the authors introduce the first evaluation metric specifically designed to assess contextual certainty adherence, systematically exposing deficiencies in current LLMs on this task. They further propose a general-purpose prompting strategy that requires no model weight modifications, integrating prior-knowledge reminders, certainty calibration, and context simplification mechanisms. Experimental results demonstrate that this approach reduces certainty adherence errors by 25% on average across multiple mainstream LLMs, substantially enhancing their ability to respond cautiously when confronted with uncertain input contexts.

0 citationsRead paper

LassoFlexNet: Flexible Neural Architecture for Tabular Data

Mar 20, 2026

Deep neural networks often underperform tree-based models on tabular data due to irrelevant features, feature heterogeneity, and local irregularities. To address this, this work proposes the LassoFlexNet architecture, which enhances model expressivity through five inductive biases: it employs per-feature embeddings to separately capture linear and nonlinear marginal contributions, integrates a grouped Lasso penalty for sparse variable selection, and introduces a novel Sequential Hierarchical Proximal Adaptive Gradient optimizer incorporating exponential moving averages (EMA). This design effectively breaks undesirable rotational invariance inherent in standard neural networks. Evaluated across 52 benchmark datasets, LassoFlexNet matches or surpasses leading tree-based models, achieving up to a 10% relative performance gain while maintaining strong interpretability.

0 citationsRead paper

On the Trade-Off Between Transparency and Security in Adversarial Machine Learning

Nov 14, 2025

This work investigates the fundamental tension between model transparency and system security in adversarial machine learning. Focusing on transferable adversarial example attacks, we formulate a game-theoretic framework—incorporating both Nash and Stackelberg equilibria—between attackers and defenders, and conduct large-scale empirical evaluation across nine attack methods and 181 models. Our analysis reveals that transparency substantially increases attack success: merely knowing whether a model is protected suffices to significantly degrade its robustness. We introduce “stealthiness” as a critical defensive advantage, formally demonstrating that strategically concealing defense mechanisms yields superior security outcomes compared to full transparency. To our knowledge, this is the first study to formalize the transparency–security trade-off using rigorous game-theoretic analysis. The findings provide both theoretical grounding and practical guidance for deploying responsible AI systems under adversarial conditions.

0 citationsRead paper