Federated Learning (FL) has emerged as a potential distributed learning paradigm that enables model training on edge devices (i.e., workers) while preserving data privacy. However, its reliance on a centralized server leads to limited scalability. Decentralized federated learning (DFL) eliminates the dependency on a centralized server by enabling peer-to peer model exchange. Existing DFL mechanisms mainly employ synchronous communication, which may result in training inefficiencies under heterogeneous and dynamic edge environments. Although a few recent asynchronous DFL (ADFL) mechanisms have been proposed to address these issues, they typically yield stale model aggregation and frequent model transmission, leading to degraded training performance on non-IID data and high communication overhead. To overcome these issues, we present DySTop, an innovative mechanism that jointly optimizes dynamic staleness control and topology construction in ADFL. In each round, multiple workers are activated, and a subset of their neighbors is selected to transmit models for aggregation, followed by local training. We provide a rigorous convergence analysis for DySTop, theoretically revealing the quantitative relationships between the convergence bound and key factors such as maximum staleness, activating frequency, and data distribution among workers. From the insights of the analysis, we propose a worker activation algorithm (WAA) for staleness control and a phase-aware topology construction algorithm (PTCA) to reduce communication overhead and handle data non-IID. Extensive evaluations through both large-scale simulations and real-world testbed experiments demonstrate that our DySTop reduces completion time by 46.7% and the communication resource consumption by 48.3% compared to state-of-the-art solutions, while maintaining the same model accuracy.

Deep neural networks (DNNs) have demonstrated exceptional performance, leading to diverse applications across various mobile devices (MDs). Considering factors like portability and environmental sustainability, an increasing number of MDs are adopting energy harvesting (EH) techniques for power supply. However, the computational intensity of DNNs presents significant challenges for their deployment on these resource-constrained devices. Existing approaches often employ DNN partition or offloading to mitigate the time and energy consumption associated with running DNNs on MDs. Nonetheless, existing methods frequently fall short in accurately modeling the execution time of DNNs, and do not consider to use thread allocation for further latency and energy consumption optimization. To solve these problems, we propose a dynamic DNN partition and thread allocation method to optimize the latency and energy consumption of running DNNs on EH-enabled MDs. Specifically, we first investigate the relationship between DNN inference latency and allocated threads and establish an accurate DNN latency prediction model. Based on the prediction model, a DRL-based DNN partition (DDP) algorithm is designed to find the optimal partitions for DNNs. A thread allocation (TA) algorithm is proposed to reduce the inference latency. Experimental results from our test-bed platform demonstrate that compared to four benchmarking methods, our scheme can reduce DNN inference latency and energy consumption by up to 37.3% and 38.5%.

Federated learning (FL) enables privacy-preserving distributed machine learning by training models on edge client devices using their local data without revealing their raw data. In edge environments, various applications require different neural network models, making it crucial to perform joint training of multiple models on edge devices, known as multi-task FL. While existing multi-task FL approaches enhance resource utilization on edge devices through adaptive resource configuration or client selection, optimizing either of these aspects alone may lead to suboptimality. Therefore, in this paper, we explore a joint client selection and resource configuration method called JCSRC for multi-task FL, aiming to maximize energy efficiency in environments with limited computation and communication resources and heterogeneous client devices. Firstly, we formalize this problem as a mixed-integer nonlinear programming problem considering all these characteristics and prove its NP-hardness. To address this problem, we first design a multi-agent reinforcement learning (MARL)-based client selection method that selects appropriate clients for each task to train their models. The MARL method makes client selection decisions based on the clients’ data quality, energy efficiency, communication, and computation capacity to ensure fast convergence and energy efficiency. Then, we design a particle swarm optimization (PSO)-based resource configuration scheme that configures appropriate computation and bandwidth resources for each task on each client. The PSO scheme makes resource configuration decisions based on theoretically derived optimal CPU frequency and bandwidth to achieve high energy efficiency. Finally, we carry out extensive simulations and testbed-based experiments to validate our proposed JCSRC. The results demonstrate that, in comparison to state-of-the-art solutions, JCSRC can save energy consumption by up to 59% to achieve the target accuracy.

While a majority of single-modal ship detectors solely rely on RGB images, a novel multi-modal real-time transformer-based ship detection and classification method, called the MM-ShipNet, is proposed in this paper that integrates the data acquired from three modalities—i.e., RGB camera, radar, and automatic identification system (AIS). First, a bounding box is generated based on the position information from radar and ship’s actual size information from AIS. This physical information are fused and projected onto the camera-acquired RGB image frame. Each bounding box is then possibly weighted depending on the ship size presented on the image. The generated weighted ship masks (WSMs) will be exploited for facilitating ship classification task. In the second stage of MM-ShipNet, multi-modal detection transformer (MM-DETR) introduces an multi-modal cross-scale encoder (MCE) for improving ship detection and classification performance. Our MCE exploits a dual-flow structure to fuse the features extracted from the WSMs and the RGB images under different scales. Since our method is the first work entailing three aforementioned modalities, no such dataset with all modalities can be found in the open source. Thus, we construct a multi-modal ship dataset, termed MMShips, as another contribution. Our MMShips dataset comprises 9,513 camera-acquired real-life maritime RGB images and their aligned ship masks generated from radar and AIS. Experimental results clearly demonstrate that our MM-ShipNet significantly outperforms multiple state-of-the-art single-modal and multi-modal ship detectors.

Molecule generation is a critical process in the fields of drug discovery and materials science. Recently, generative models based on normalizing flows have demonstrated significant potential in this domain. These models are particularly suited for handling the symmetrical and complex chemical structures often encountered in molecular datasets. Despite their promising nature, normalizing flow-based models for molecule generation face considerable challenges. The complexity of molecule representation, the rigorous demands of optimization, and the scarcity of training labels in molecular datasets contribute to these difficulties. Additionally, adequately and comprehensively learning the distribution of molecular datasets remains a formidable task. In this paper, we delve into the intricate entry-wise modules in vanilla flows, introducing an effective variation of flow-based models. Our proposed approach innovatively encapsulates affine coupling transformations within normalizing flows. Furthermore, we deconstruct existing invertible flow models, integrating them with newly developed entry-wise transformations. Our experimental studies demonstrate that these proposed entry-wise modules, when incorporated into standard flow-based models, surpass other generative models in performance on various representative datasets and generation tasks. Notably, in the context of low-resourced molecular graph generation, our model achieves remarkable performance compared to its counterparts.