Selected work

Representative Papers

Cross-Ecosystem Bug Classification in Quantum Software

Aug 04, 2026

This study addresses the complex and poorly understood defects arising from interactions between classical and quantum components in quantum software, where existing research lacks a unified, cross-ecosystem classification methodology. The work proposes the first rule-driven, interpretable, and automatable classification framework, systematically annotating and comparing 17,523 issues across 12 repositories—including Qiskit, Cirq, and PyQuil—along dimensions of defect type, severity, quality attributes, and quantum-specific subcategories. Through statistical validation, machine learning baselines, and longitudinal trend analysis (2017–2025), the study reveals that classical defects constitute 67% of all issues, while quantum-specific defects remain stable at 27–30%, with significant ecosystem-level variations (e.g., prominent compatibility issues in Qiskit). The proposed framework substantially outperforms data-driven models in fine-grained quantum defect identification.

0 citationsRead paper

Fairness-Aware Profit Maximization using Deep Reinforcement Learning

May 28, 2026

This study addresses the problem of profit maximization under budget constraints in social networks while ensuring fairness across communities. Specifically, it requires that each community receives a minimum guaranteed profit according to the maximin fairness criterion when selecting seed nodes. To this end, the work introduces maximin fairness constraints into the influence propagation–based profit maximization framework for the first time and formulates the problem as a Markov decision process, which is efficiently solved using a deep Q-network. Experimental results on real-world social network datasets demonstrate that the proposed method significantly outperforms existing baselines, achieving up to ten times higher profit while effectively maintaining group-level fairness.

0 citationsRead paper

Approximation Algorithms for Budget Splitting in Multi-Channel Influence Maximization

Apr 01, 2026

This study addresses the budget allocation problem across multiple advertising channels—such as billboards and social networks—and presents the first formal model capturing inter-channel interaction effects on overall influence. The resulting influence function is non-submodular, prompting the development of an approximation framework based on the submodularity ratio and generalized curvature. The authors propose a randomized greedy algorithm and a two-stage adaptive strategy to optimize allocations under this setting. They establish the first theoretical approximation guarantee for such non-submodular influence maximization, achieving an approximation ratio of $\frac{1}{\alpha}(1 - e^{-\gamma \alpha})$. Extensive experiments on real-world datasets demonstrate that the proposed methods significantly outperform existing approaches in enhancing total influence.

0 citationsRead paper

Tag-specific Regret Minimization Problem in Outdoor Advertising

Mar 06, 2026

This study addresses the label assignment problem in outdoor advertising, aiming to minimize total regret—which is non-monotonic and non-submodular—while satisfying advertisers’ influence requirements and budget constraints. To tackle this NP-hard and inapproximable problem, the authors propose a combinatorial optimization model termed TRMOA and develop a fairness-aware greedy round-robin strategy that integrates randomized greedy selection with local search. Experimental evaluations on real-world user trajectory and billboard datasets demonstrate that the proposed approach significantly reduces regret and enhances allocation efficiency, effectively balancing fairness and computational tractability.

0 citationsRead paper

Generalizable Federated Learning using Client Adaptive Focal Modulation

Aug 14, 2025

To address the weak generalization of federated learning (FL) under non-IID and cross-domain settings, as well as the inherent trade-off between privacy preservation and communication efficiency, this paper proposes a client-adaptive focal modulation framework tailored for multimodal, resource-constrained environments. Methodologically, it introduces: (1) a task-aware client embedding mechanism that dynamically generates personalized modulation strategies; (2) a lightweight modulation layer generation approach based on low-rank hypernetworks, drastically reducing communication overhead; and (3) a unified training paradigm integrating Transformer architectures with cross-modal adaptive optimization. Extensive experiments across eight benchmark datasets demonstrate that our method significantly outperforms state-of-the-art approaches in both source-free federated and cross-task scenarios. It achieves superior generalization capability, high communication efficiency, and strong scalability—without compromising model utility or privacy guarantees.

0 citationsRead paper

Recent publications

Latest Papers

Cross-Ecosystem Bug Classification in Quantum Software

Aug 04, 2026

This study addresses the complex and poorly understood defects arising from interactions between classical and quantum components in quantum software, where existing research lacks a unified, cross-ecosystem classification methodology. The work proposes the first rule-driven, interpretable, and automatable classification framework, systematically annotating and comparing 17,523 issues across 12 repositories—including Qiskit, Cirq, and PyQuil—along dimensions of defect type, severity, quality attributes, and quantum-specific subcategories. Through statistical validation, machine learning baselines, and longitudinal trend analysis (2017–2025), the study reveals that classical defects constitute 67% of all issues, while quantum-specific defects remain stable at 27–30%, with significant ecosystem-level variations (e.g., prominent compatibility issues in Qiskit). The proposed framework substantially outperforms data-driven models in fine-grained quantum defect identification.

0 citationsRead paper

Fairness-Aware Profit Maximization using Deep Reinforcement Learning

May 28, 2026

This study addresses the problem of profit maximization under budget constraints in social networks while ensuring fairness across communities. Specifically, it requires that each community receives a minimum guaranteed profit according to the maximin fairness criterion when selecting seed nodes. To this end, the work introduces maximin fairness constraints into the influence propagation–based profit maximization framework for the first time and formulates the problem as a Markov decision process, which is efficiently solved using a deep Q-network. Experimental results on real-world social network datasets demonstrate that the proposed method significantly outperforms existing baselines, achieving up to ten times higher profit while effectively maintaining group-level fairness.

0 citationsRead paper

Approximation Algorithms for Budget Splitting in Multi-Channel Influence Maximization

Apr 01, 2026

This study addresses the budget allocation problem across multiple advertising channels—such as billboards and social networks—and presents the first formal model capturing inter-channel interaction effects on overall influence. The resulting influence function is non-submodular, prompting the development of an approximation framework based on the submodularity ratio and generalized curvature. The authors propose a randomized greedy algorithm and a two-stage adaptive strategy to optimize allocations under this setting. They establish the first theoretical approximation guarantee for such non-submodular influence maximization, achieving an approximation ratio of $\frac{1}{\alpha}(1 - e^{-\gamma \alpha})$. Extensive experiments on real-world datasets demonstrate that the proposed methods significantly outperform existing approaches in enhancing total influence.

0 citationsRead paper

Tag-specific Regret Minimization Problem in Outdoor Advertising

Mar 06, 2026

This study addresses the label assignment problem in outdoor advertising, aiming to minimize total regret—which is non-monotonic and non-submodular—while satisfying advertisers’ influence requirements and budget constraints. To tackle this NP-hard and inapproximable problem, the authors propose a combinatorial optimization model termed TRMOA and develop a fairness-aware greedy round-robin strategy that integrates randomized greedy selection with local search. Experimental evaluations on real-world user trajectory and billboard datasets demonstrate that the proposed approach significantly reduces regret and enhances allocation efficiency, effectively balancing fairness and computational tractability.

0 citationsRead paper

Generalizable Federated Learning using Client Adaptive Focal Modulation

Aug 14, 2025

To address the weak generalization of federated learning (FL) under non-IID and cross-domain settings, as well as the inherent trade-off between privacy preservation and communication efficiency, this paper proposes a client-adaptive focal modulation framework tailored for multimodal, resource-constrained environments. Methodologically, it introduces: (1) a task-aware client embedding mechanism that dynamically generates personalized modulation strategies; (2) a lightweight modulation layer generation approach based on low-rank hypernetworks, drastically reducing communication overhead; and (3) a unified training paradigm integrating Transformer architectures with cross-modal adaptive optimization. Extensive experiments across eight benchmark datasets demonstrate that our method significantly outperforms state-of-the-art approaches in both source-free federated and cross-task scenarios. It achieves superior generalization capability, high communication efficiency, and strong scalability—without compromising model utility or privacy guarantees.

0 citationsRead paper