TLTan Li

Tan Li

*This is an algorithmically generated author record based on publicly available information.

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


Tan Li has a total of 34 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Adaptive Layer-Wise Personalized Federated Deep Reinforcement Learning for Heterogeneous Edge Caching

    by Tan Li, Zhen Li, Hai Liu, Chao Yang, Tse-Tin Chan, Jun Cai

    Proactive caching is essential for minimizing latency and improving Quality of Experience (QoE) in heterogeneous edge networks. While Federated Deep Reinforcement Learning (FDRL) shows promise for developing cache policies, it faces challenges such as an expanding action space and difficulty in balancing global knowledge sharing with local environmental adaptation. In this paper, we propose a Layer-wise Relevance Propagation-aided Personalized Federated (LRP-PFed) Deep Reinforcement Learning framework for edge caching to maximize system utility while satisfying caching constraints. To handle the expanding action space, we design a Multi-Head Double Deep Q-Network (MH-DDQN) that reshapes the action output layers into a multi-head structure, where each head generates a sub-dimensional action. Furthermore, we introduce an LRP-based adaptive personalization mechanism that dynamically determines the optimal number of personalized layers for each edge server during training. This approach enables automatic adaptation to heterogeneous environments while leveraging global information to accelerate learning convergence. Extensive experiments validate the effectiveness of our approach, showing that MH-DDQN achieves superior cache hit rates and reduced computational complexity compared to traditional DRL methods, while our LRP-guided personalization strategy achieves superior performance, scalability, and adaptivity compared to existing FDRL methods.


  • Client Association and Resource Allocation for NOMA-Enabled Hierarchical Federated Learning With Non-IID Data

    by Tan Li, Linqi Song

    Hierarchical federated learning (HFL) provides a new client-edge-cloud paradigm to reduce the network burden brought by centralized FL. However, the current HFL framework mainly adopts orthogonal multiple access (OMA) techniques for client-edge communication, failing to cope with massive connections with limited bandwidth resources. Besides, the highly skewed data distribution across the clients can undermine the convergence, which leads to more rounds to reach the target performance. In this paper, we consider a non-orthogonal multiple access (NOMA)-enabled HFL framework that provides a spectrum-efficient approach to enable massive client connectivity. Under this framework, we investigate the joint client-edge association and resource allocation problem to optimize the dual-efficiency. We first reformulate the original problem as a bi-objective optimization problem using dynamic weights to balance the resource and data heterogeneity. After that, we formulate the edge-client association problem as a many-to-one matching game with externalities and swap the connected pairs until a stable matching is obtained. For each temporary edge-client association, we solve the CPU frequency and transmission power control problem for all clients using convex optimization methods. We conduct experiments under various unbalanced data distributions and communication environment settings. The results demonstrate that our method outperforms other benchmarks regarding client scheduling, transmission and resource allocation.


  • Layered Randomized Quantization for Communication-Efficient and Privacy-Preserving Distributed Learning

    by Guangfeng Yan, Tan Li, Kui Wu, Linqi Song

    In distributed learning systems, ensuring efficient communication and privacy protection are two significant challenges. Although several existing works have attempted to address these challenges simultaneously, they often overlook essential learning-oriented features such as dynamic gradient and communication characteristics. In this paper, we propose a communication-efficient and privacy-preserving distributed SGD algorithm. Our proposed algorithm employs a layered randomized quantizer (LRQ) to reduce communication overhead, which also ensures that quantization errors follow an exact Gaussian distribution, thus achieving client-level differential privacy. We analyze the trade-off between convergence error, communication, and privacy under non-IID data distributions. Besides, we modify the algorithm to be training-adaptive by adjusting the per-round privacy budget allocation in response to i) dynamic gradient features and ii) real-time changing communication rounds. Both closed-form solutions are derived by solving the minimization problem of convergence error subject to the privacy budget constraint. Finally, we evaluate the effectiveness of our approach through extensive experiments on various datasets, including MNIST, CIFAR-10, and CIFAR-100, demonstrating its superiority in terms of communication cost, privacy protection, and model performance compared to state-of-the-art methods.


  • Unified Multi-Modal Diagnostic Framework With Reconstruction Pre-Training and Heterogeneity-Combat Tuning

    by Yupei Zhang, Li Pan, Qiushi Yang, Tan Li, Zhen Chen

    Medical multi-modal pre-training has revealed promise in computer-aided diagnosis by leveraging large-scale unlabeled datasets. However, existing methods based on masked autoencoders mainly rely on data-level reconstruction tasks, but lack high-level semantic information. Furthermore, two significant heterogeneity challenges hinder the transfer of pre-trained knowledge to downstream tasks, i.e. , the distribution heterogeneity between pre-training data and downstream data, and the modality heterogeneity within downstream data. To address these challenges, we propose a Unified Medical Multi-modal Diagnostic (UMD) framework with tailored pre-training and downstream tuning strategies. Specifically, to enhance the representation abilities of vision and language encoders, we propose the Multi-level Reconstruction Pre-training (MR-Pretrain) strategy, including a feature-level and data-level reconstruction, which guides models to capture the semantic information from masked inputs of different modalities. Moreover, to tackle two kinds of heterogeneities during the downstream tuning, we present the heterogeneity-combat downstream tuning strategy, which consists of a Task-oriented Distribution Calibration (TD-Calib) and a Gradient-guided Modality Coordination (GM-Coord). In particular, TD-Calib fine-tunes the pre-trained model regarding the distribution of downstream datasets, and GM-Coord adjusts the gradient weights according to the dynamic optimization status of different modalities. Extensive experiments on five public medical datasets demonstrate the effectiveness of our UMD framework, which remarkably outperforms existing approaches on three kinds of downstream tasks.


  • Open Access

    Layered Randomized Quantization for Communication-Efficient and Privacy-Preserving Distributed Learning

    by Guangfeng Yan, Tan Li, Tian Lan, Kui Wu, Linqi Song

    Next-generation wireless networks, such as edge intelligence and wireless distributed learning, face two critical challenges: communication efficiency and privacy protection. In this work, our focus is on addressing these issues in a distributed learning framework. We consider a new approach that simultaneously achieves communication efficiency and privacy protection by exploiting the privacy advantage offered by quantization. Specifically, we use a quantization scheme called \textbf{Gau}ssian \textbf{L}ayered \textbf{R}andomized \textbf{Q}uantization (Gau-LRQ) that compresses the raw model gradients using a layer multishift coupler. By adjusting the parameters of Gau-LRQ, we shape the quantization error to follow the expected Gaussian distribution, thus ensuring client-level differential privacy (CLDP). We demonstrate the effectiveness of our proposed Gau-LRQ in the distributed stochastic gradient descent (SGD) framework and theoretically quantify the trade-offs between communication, privacy, and convergence performance. We further improve the convergence performance by enabling dynamic private budget and quantization bit allocation. We achieve this by using an optimization formula that minimizes convergence error subject to the privacy budget constraint. We evaluate our approach on multiple datasets, including MNIST, CIFAR-10, and CIFAR-100, and show that our proposed method outperforms the baselines in terms of learning performance under various privacy constraints. Moreover, we observe that dynamic privacy allocation yields additional accuracy improvements for the models compared to the fixed scheme.

  • Federated Online Learning Aided Multi-Objective Proactive Caching in Heterogeneous Edge Networks

    by Tan Li, Linqi Song

    To address the drastic increase in multimedia traffic volume, mobile edge caching (MEC) has been exploited to reduce redundant data transmissions by equipping computation and storage capacity at the edge network. Previous works on learning-based caching problems often only concern pre-storing popular contents to satisfy users’ demands. In this work, we investigate the cache strategy design problem with two possibly conflicting objectives, namely, cache hit and cache profit, in heterogeneous multi-MEC server networks when content profiles are unknown. We then formulate this multi-objective caching problem as a Multi-agent Multi-objective Combinatorial Multi-Armed bandit (MMC-MAB) problem and propose a two-step caching framework that estimates content properties first and then optimizes cache placement. Specifically, to accommodate the system heterogeneity in estimation, we utilize an adaptive federated learning-based estimation to approach the unknown content popularity and profit profiles, which wisely use both local and external observations by adjusting the mixing factors. To address the multiple objective optimizations, we propose two effective methods, based on individual dominance and combinatorial dominance, to achieve adequate Pareto-optimal cache placement. Both theoretical results and comprehensive experiments clearly validate the effectiveness and efficiency of our proposed approaches.


  • AC-SGD: Adaptively Compressed SGD for Communication-Efficient Distributed Learning

    by Guangfeng Yan, Tan Li, Shao-Lun Huang, Tian Lan, Linqi Song

    Gradient compression (e.g., gradient quantization and gradient sparsification) is a core technique in reducing communication costs in distributed learning systems. The recent trend of gradient compression is to use a varying number of bits across iterations, however, relying on empirical observations or engineering heuristics without a systematic treatment and analysis. To the best of our knowledge, a general dynamic gradient compression that leverages both quantization and sparsification techniques is still far from understanding. This paper proposes a novel Adaptively-Compressed Stochastic Gradient Descent (AC-SGD) strategy to adjust the number of quantization bits and the sparsification size with respect to the norm of gradients, the communication budget, and the remaining number of iterations. In particular, we derive an upper bound, tight in some cases, of the convergence error for arbitrary dynamic compression strategy. Then we consider communication budget constraints and propose an optimization formulation - denoted as the Adaptive Compression Problem (ACP) - for minimizing the deep model’s convergence error under such constraints. By solving the ACP, we obtain an enhanced compression algorithm that significantly improves model accuracy under given communication budget constraints. Finally, through extensive experiments on computer vision and natural language processing tasks on MNIST, CIFAR-10, CIFAR-100 and AG-News datasets, respectively, we demonstrate that our compression scheme significantly outperforms the state-of-the-art gradient compression methods in terms of mitigating communication costs.


  • Privacy-Preserving Communication-Efficient Federated Multi-Armed Bandits

    by Tan Li, Linqi Song

    Communication bottleneck and data privacy are two critical concerns in federated multi-armed bandit (MAB) problems, such as situations in decision-making and recommendations of connected vehicles via wireless. In this paper, we design the privacy-preserving communication-efficient algorithm in such problems and study the interactions among privacy, communication and learning performance in terms of the regret. To be specific, we design privacy-preserving learning algorithms and communication protocols and derive the learning regret when networked private agents are performing online bandit learning in a master-worker, a decentralized and a hybrid structure. Our bandit learning algorithms are based on epoch-wise sub-optimal arm eliminations at each agent and agents exchange learning knowledge with the server/each other at the end of each epoch. Furthermore, we adopt the differential privacy (DP) approach to protect the data privacy at each agent when exchanging information; and we curtail communication costs by making less frequent communications with fewer agents participation. By analyzing the regret of our proposed algorithmic framework in the master-worker, decentralized and hybrid structures, we theoretically show trade-offs between regret and communication costs/privacy. Finally, we empirically show these trade-offs which are consistent with our theoretical analysis.