8 March 2024 in IEEE Transactions on Mobile Computing
Recently, federated learning (FL) as a new learning paradigm allows multi-party to collaboratively train a shared global model with privacy protection. However, vanilla FL running on heterogeneous mobile edge devices still faces three crucial challenges: communication efficiency, statistical heterogeneity, and system heterogeneity. To tackle them simultaneously, we devise FedPE , a communication-efficient and personalized federated learning framework, which allows each client to search for personalized optimal local subnets adaptive to system capacity in each round of FL. It consists of three core components: a) adaptive pruning-expanding controls model pruning or expanding according to the accuracy variations of local models, b) error compensation strategy promotes the pruned or expanded subnets to be Lottery Ticket Networks (LTNs), c) the fair aggregation rule aggregates local models with their real-time contributions as coefficients to boost the performance of the aggregated global model. The integration of the three components facilitates that only personalized optimal subnets with different footprints interact between the server and clients, which effectively reduces communication costs and enhances the robustness of FL to statistical and system heterogeneity. We also prove the convergence of FedPE and design an optimal hyperparameter searching (OHS) algorithm based on Pareto optimization to search for optimal hyperparameters for FedPE . Extensive experiments evaluated on five real-world datasets with IID or Non-IID distributions demonstrate that FedPE configured with found optimal hyperparameters achieves $1.86\times -121\times$ communication efficiency improvement with almost no accuracy degradation, presenting the best trade-off between model accuracy and communication cost.
14 November 2022 in IEEE Transactions on Circuits and Systems for Video Technology
Recent online knowledge distillation (OKD) methods focus on capturing rich and useful intermediate information by performing multi-layer feature learning. Existing works only consider intermediate layer feature maps between the same layers and ignore valuable information across layers, which results in the lack of appropriate cross-layer supervision in detail and the process of learning. Besides, this manner provides insufficient supervision information to supervise the learning of student, since it fails to construct a qualified teacher. In this work, we propose a Deep Cross-layer Collaborative Learning network (DCCL) for OKD, which efficiently exploits fruitful knowledge of peer student models by keeping appropriate intermediate cross-layer supervision. Specifically, each student gradually integrates its own features at different layers for feature matching, so as to effectively utilize features in low and high levels for learning more composite knowledge. Moreover, we assign a collaborative knowledge learning strategy, in which a qualified teacher is established via fusing the features of last convolution layers for enhancing high-level representation. In this way, all student models continuously transfer the rich teacher’s internal representation as well as capture its dynamic growth process, and in turn assist the learning of the fusion teacher to further supervise students. In the experiments, our proposed DCCL has shown great generalization ability with various backbone models on CIFAR-100, Tiny ImageNet and ImageNet, and also demonstrated superior performance against mainstream OKD works. Our code is available here: https://github.com/nanxiaotong/DCCL.
14 April 2022 in IEEE Transactions on Neural Networks and Learning Systems
Knowledge distillation (KD) transfers discriminative knowledge from a large and complex model (known as teacher) to a smaller and faster one (known as student). Existing advanced KD methods, limited to fixed feature extraction paradigms that capture teacher's structure knowledge to guide the training of the student, often fail to obtain comprehensive knowledge to the student. Toward this end, in this article, we propose a new approach, synchronous teaching knowledge distillation (STKD), to integrate online teaching and offline teaching for transferring rich and comprehensive knowledge to the student. In the online learning stage, a blockwise unit is designed to distill the intermediate-level knowledge and high-level knowledge, which can achieve bidirectional guidance of the teacher and student networks. Intermediate-level information interaction provides more supervisory information to the student network and is useful to enhance the quality of final predictions. In the offline learning stage, the STKD approach applies a pretrained teacher to further improve the performance and accelerate the training process by providing prior knowledge. Trained simultaneously, the student learns multilevel and comprehensive knowledge by incorporating online teaching and offline teaching, which combines the advantages of different KD strategies through our STKD method. Experimental results on the SVHN, CIFAR-10, CIFAR-100, and ImageNet ILSVRC 2012 real-world datasets show that the proposed method achieves significant performance improvements compared with the state-of-the-art methods, especially with satisfying accuracy and model size. Code for STKD is provided at https://github.com/nanxiaotong/STKD.
27 June 2025 in Expert Systems with Applications