Jingyang Zhu

Jingyang Zhu, Yuanming Shi, Yong Zhou et al.

Federated learning (FL), as an emerging distributed machine learning paradigm, allows a mass of edge devices to collaboratively train a global model while preserving privacy. In this tutorial, we focus on FL via over-the-air computation (AirComp), which is proposed to reduce the communication overhead for FL over wireless networks at the cost of compromising in the learning performance due to model aggregation error arising from channel fading and noise. We first provide a comprehensive study on the convergence of AirComp-based FedAvg (AirFedAvg) algorithms under both strongly convex and non-convex settings with constant and diminishing learning rates in the presence of data heterogeneity. Through convergence and asymptotic analysis, we characterize the impact of aggregation error on the convergence bound and provide insights for system design with convergence guarantees. Then we derive convergence rates for AirFedAvg algorithms for strongly convex and non-convex objectives. For different types of local updates that can be transmitted by edge devices (i.e., local model, gradient, and model difference), we reveal that transmitting local model in AirFedAvg may cause divergence in the training procedure. In addition, we consider more practical signal processing schemes to improve the communication efficiency and further extend the convergence analysis to different forms of model aggregation error caused by these signal processing schemes. Extensive simulation results under different settings of objective functions, transmitted local information, and communication schemes verify the theoretical conclusions.

Yuanming Shi, Jingyang Zhu, Chunxiao Jiang et al.

Driven by the growing demand for intelligent remote sensing applications, large artificial intelligence (AI) models pre-trained on large-scale unlabeled datasets and fine-tuned for downstream tasks have significantly improved learning performance for various downstream tasks due to their generalization capabilities. However, many specific downstream tasks, such as extreme weather nowcasting (e.g., downburst and tornado), disaster monitoring, and battlefield surveillance, require real-time data processing. Traditional methods via transferring raw data to ground stations for processing often cause significant issues in terms of latency and trustworthiness. To address these challenges, satellite edge AI provides a paradigm shift from ground-based to on-board data processing by leveraging the integrated communication-and-computation capabilities in space computing power networks (Space-CPN), thereby enhancing the timeliness, effectiveness, and trustworthiness for remote sensing downstream tasks. Moreover, satellite edge large AI model (LAM) involves both the training (i.e., fine-tuning) and inference phases, where a key challenge lies in developing computation task decomposition principles to support scalable LAM deployment in resource-constrained space networks with time-varying topologies. In this article, we first propose a satellite federated fine-tuning architecture to split and deploy the modules of LAM over space and ground networks for efficient LAM fine-tuning. We then introduce a microservice-empowered satellite edge LAM inference architecture that virtualizes LAM components into lightweight microservices tailored for multi-task multimodal inference. Finally, we discuss the future directions for enhancing the efficiency and scalability of satellite edge LAM, including task-oriented communication, brain-inspired computing, and satellite edge AI network optimization.

Xizi Chen, Jingyang Zhu, Jingbo Jiang et al.

The unstructured sparsity after pruning poses a challenge to the efficient implementation of deep learning models in existing regular architectures like systolic arrays. On the other hand, coarse-grained structured pruning is suitable for implementation in regular architectures but tends to have higher accuracy loss than unstructured pruning when the pruned models are of the same size. In this work, we propose a model compression method based on a novel weight permutation scheme to fully exploit the fine-grained weight sparsity in the hardware design. Through permutation, the optimal arrangement of the weight matrix is obtained, and the sparse weight matrix is further compressed to a small and dense format to make full use of the hardware resources. Two pruning granularities are explored. In addition to the unstructured weight pruning, we also propose a more fine-grained subword-level pruning to further improve the compression performance. Compared to the state-of-the-art works, the matrix compression rate is significantly improved from 5.88x to 14.13x. As a result, the throughput and energy efficiency are improved by 2.75 and 1.86 times, respectively.