Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Qingguo Lü has a total of 106 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Decentralized online stochastic dual averaging algorithm for convex nonseparable problems

    by Xiaohui Zhou, Huqiang Cheng, Qingguo Lü, Huaqing Li, Xiaofeng Liao

  • Distributed Online Learning for Constrained Nonseparable Problems With Privacy Preservation

    by Huqiang Cheng, Huaqing Li, Jinhui Hu, Qingguo Lü, You Zhao, Tingwen Huang

    In large-scale distributed systems, data is stored across multiple locations and must be processed in real time to ensure rapid responses to consumer demands. However, this distributed nature introduces significant privacy challenges, as the data often contains sensitive information such as salaries, medical records, or personal identifiers. Considering these factors, this work centers on developing differentially private distributed online algorithms. The goal is to tackle the constrained online convex nonseparable problems over unbalanced directed networks. To this end, we first design an algorithmic framework by leveraging Nesterov’s dual averaging technique, where every node (learner) is only responsible for training a part of the model parameter. Under this framework, we develop a differentially private distributed online dual averaging learning algorithm, termed DP-DARE, by incorporating the stochastic subgradient rescaling technique. We demonstrate that DP-DARE achieves sublinear expected dynamic regret under various Laplacian noise scale parameters (increasing, constant, and decreasing) while ensuring $\epsilon$-differential privacy across all scenarios. Moreover, the robustness of DP-DARE to arbitrary, fixed, but bounded communication delays is briefly discussed. Experimental evaluations ultimately confirm the validity of the analytical findings.

  • Open Access

    LOIRP: Convex Optimization and Mahalanobis Filtering for Noisy Imbalanced Classification

    by Jingxin Liu, Fan Shu, Chunhong Du, Biao Chen, Hao Zhou, Witold Pedrycz, Jingwen Tu, Qingguo Lu

    Class imbalance and label noise coexist in many real-world applications, and the classifier tends to exhibit systematic bias towards the majority class. The Existing oversampling methods rely on labelconsistency signals or fixed sampling ratios, both of which degrade under label corruption. LOIRP is proposed as a lightweight two-module framework that optimizes synthetic sample quantity and quality via label-independent criteria. Module I designs a strictly convex objective encoding the imbalance ratio and covariance structure, which is solved via discrete grid search with a 𝜏-bound mechanism guaranteeing minimum minority-class representation. Module II applies a Mahalanobis surrogate filter that scores candidates by the ratio of inter-class Euclidean distance to intra-class Mahalanobis distance, thereby enabling robust quality assessment under arbitrary noise rates. The extensive experiments on 28 real-world datasets spanning nine domains, four classifiers, and threenoise levels demonstrate that LOIRP achieves the best F1, AUC-PR, and MCC under moderate and severe noise, with statistically significant improvements confirmed by Friedman and Wilcoxon tests. Compared with the optimization baseline OIRP, LOIRP achieves competitive classification quality with an average speedup of approximately 2 x 10

  • Open Access

    OURF: Optimization-driven undersampling robust framework for multi-class imbalanced noisy learning

    by Min Li, Hao Zhou, Qun Liu, Qingguo Lu, Huaqing Li, Ponnuthurai Nagaratnam Suganthan

    Multi-class imbalanced noisy learning is fundamentally challenged by the intricate interplay among class imbalance, structural degradation, and label noise. Existing undersampling methods predominantly rely on fixed-ratio reduction or heuristic sample removal, often leading to information collapse, manifold distortion, and deteriorated decision boundaries while remaining vulnerable to noise propagation. To address these limitations without generating additional minority samples, this paper proposes an optimization-driven undersampling robust framework (OURF) for multi-class imbalanced noisy learning. Unlike conventional undersampling that focus primarily on numerical class balancing, OURF reinterprets undersampling as a structural-risk control problem and, for the first time, establishes a two-stage optimization framework from the perspective of optimal balance, enabling globally coordinated optimization among class balancing, structure preservation, and noise suppression. Specifically, a structure-risk balanced undersampling quantity optimization (SRB-UQO) model is developed to adaptively determine class-wise retention quantities through a structure-noise adaptive target prior, thereby preventing information collapse caused by excessive compression. Building upon the learned quantity constraints, a noise-robust manifold-preserving sample–feature optimization (NRMP-SFO) model further performs structure-robust representation refinement, producing retained subsets that preserve geometric consistency, topological integrity, and discriminative stability. From a theoretical perspective, we establish a unified structural-risk control framework for optimization-driven undersampling. The analysis reveals that the fundamental limitation of conventional undersampling stems from its inability to jointly control coupled structural risks, including information collapse, manifold fragmentation, and noise exposure. Furthermore, we prove that OURF tightens a provable generalization-risk upper bound through the joint optimization of multiple computable structural-risk surrogates. Extensive experiments demonstrate that OURF serves as a general optimization framework that not only consistently enhances representative undersampling methods but also yields a competitive standalone undersampling scheme. Across diverse noise scenarios, OURF exhibits superior structural robustness, stability, and cross-noise generalization capability compared with SOTA methods.

  • Decentralized Constrained Optimization Over Time-Varying Directed Networks via Subgradient Rescaling

    by Qingguo Lü, Huaqing Li, Chaoxu Wu, Hao Zhou, Tingwen Huang, Ponnuthurai Nagaratnam Suganthan

    In this article, we investigate a decentralized constrained optimization problem over time-varying directed networks. The nodes in the network aim to collaboratively minimize the aggregate of all locally known convex cost functions, subject to local nonidentical and multiple constraint sets, inequality constraints, and equality constraints. Problems of this nature arise in a number of applications in real networks, such as facility location in wireless sensor networks and image deblurring in machine learning. To address these types of problems, we propose an efficient subgradient-rescaling-based decentralized fixed-random projection (SR-DFRP) algorithm, named the SR-DFRP algorithm. In particular, the SR-DFRP algorithm employs Polyak's random projection to handle nonidentical and multiple constraints, which reduces computational load by avoiding the formulation of complex subproblems. Furthermore, by utilizing dynamically constructed row-stochastic matrices, the algorithm employs a subgradient rescaling strategy to mitigate the imbalance induced by the time-varying directed networks. Rigorous theoretical analyses are provided to establish that the SR-DFRP algorithm converges almost surely to the optimal solution. Extensive simulations on facility location and image deblurring problems are presented to validate the efficacy of the algorithm and the validity of the theoretical results.


  • NO-IGS: Nonlinear Oversampling Based on Informed Granular-ball Space for Imbalanced Regression

    by Hao Zhou, Xiaofeng Liao, Qingguo Lu, Ponnuthurai Nagaratnam Suganthan

    In this paper, we consider the problem of imbalanced regression. This problem is prevalent in real-world regression datasets. Resampling is a critical technique utilized in imbalanced regression, which simply and directly manipulates the quantity of skewed samples to rebalance the data distribution. However, the relevant academic research mainly focused on classification, with little discussion on imbalanced regression. Therefore, an adaptive Nonlinear Oversampling within Informed Granular-ball Space (NO-IGS) is proposed for imbalanced regression as a novel direction. First, NO-IGS proposes a nonconvex optimization model to fit the distribution of the dataset and adaptively split the granular-ball space. Then, NO-IGS proposes another non-convex optimization model to select an appropriate granular-ball space for oversampling. The existence of optimal solutions for both optimization models is proven, and solution algorithms are provided. Finally, based on anisotropic multivariate Gaussian distributions, new samples conforming to the original distribution of the dataset are synthesized in the granular-ball space. Numerical experiments based on dozens of datasets and comparison algorithms demonstrate the effectiveness and robustness of NO-IGS.


  • Digitally Implementable Dynamic Output Control for Multiagent Systems With DoS Attacks and Disturbances via Multiple Types of Triggered Mechanisms

    by Yonghui Wu, Huaqing Li, Huiwei Wang, Qingguo Lü

    This study addresses the cooperative control of multiagent systems (MASs) under disturbances and denial-of-service (DoS) attacks. To align with digital hardware implementation and eliminate reliance on a global coordinate frame, a time-triggered mechanism (TTM) and a hybrid time-dynamic event-triggered mechanism (ETM) are proposed for sampling relative output measurements and coordinating interagent communication. To avoid continuous monitoring and computation, a periodic event-triggered scheme is adopted, restricting these processes to predefined-time sequences. Under verifiable conditions, the existence of suitable control parameters is proven through a parameter selection procedure. The relationship between parameter choice, resilience to DoS attacks, and the convergence error under disturbances is also analyzed, ensuring predictable control performance. Numerical simulations are provided to demonstrate the effectiveness of the proposed approach.

  • A Distributed Stochastic Unified Accelerated Algorithm for Convex Optimization Over Time-Varying Directed Networks

    by Bingxue Luo, Qingguo Lü, Huqiang Cheng, Hao Zhou, Xiaofeng Liao

    In the current era, distributed optimization over multinode systems has become the mainstream approach for tackling large-scale complex problems. However, a key challenge is that the data available for distributed optimization in practical engineering environments is often inherently noisy or inaccurate. This problem turns into more complicated when both noise-tolerant and accelerated convergence are desired. To address these problems, this article investigates a distributed stochastic optimization problem over time-varying directed networks where all nodes work together to minimize the sum of local cost functions. Specially, under the condition that each node of the network has the ability to acquire the stochastic first-order oracle ( $\mathcal {S} \mathcal {F}\mathcal {O}$ ), we design an efficient distributed stochastic unified accelerated algorithm, named DSUA-TV. First, DSUA-TV acquires an unbiased gradient assessment of its local cost function by accessing $\mathcal {S} \mathcal {F}\mathcal {O}$ , and the global variance arising from the diversity of nodes’ information is mitigated by the gradient tracking mechanism. Second, DSUA-TV focuses on time-varying directed networks and incorporates a unified momentum acceleration framework, providing additional improvements in both convergence speed and stability. Finally, it proves that DSUA-TV can converge linearly to an error ball around the global optimal solution in expectation, as long as the strong convexity and smoothness of nodes’ cost functions are ensured. Moreover, simulation results further exhibit the high efficiency of DSUA-TV.