Data-Free Knowledge Distillation for Heterogeneous Federated Learning

By Zhuangdi Zhu, Junyuan Hong, Jiayu Zhou. International Conference on Machine Learning, 2021

1,012 citations recorded by the source corpus.

Federated Learning (FL) is a decentralized machine-learning paradigm in which a global server iteratively aggregates the model parameters of local users without accessing their data. User heterogeneity has imposed significant challenges to FL, which can incur drifted global models that are slow to converge. Knowledge Distillation has recently emerged to tackle this issue, by refining the server model using aggregated knowledge from heterogeneous users, other than directly aggregating their model parameters. This approach, however, depends on a proxy dataset, making it impractical unless such prerequisite is satisfied. Moreover, the ensemble knowledge is not fully utilized to guide local model learning, which may in turn affect the quality of the aggregated model. In this work, we propose a data-free knowledge distillation approach to address heterogeneous FL, where the server learns a lightweight generator to ensemble user information in a data-free manner, which is then broadcasted to users, regulating local training using the learned knowledge as an inductive bias. Empirical studies powered by theoretical implications show that, our approach facilitates FL with better generalization performance using fewer communication rounds, compared with the state-of-the-art1.

Subfields: Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/3fdcf2d8fe58dcbc2c353e5974e2030f8281c799

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Dacheng Tao
2. Bo Du
3. Qiang Yang
4. Han Yu
5. Mang Ye
6. Pong Chi Yuen
7. Li-zhen Cui
8. Alysa Ziying Tan

Citations in the corpus

• Towards Personalized Federated Learning (2021, methodology): “As such, there are PFL works that use distribution-based evaluation frameworks such as histogram profiling [61], [87], variance metrics [37], [55], [88] and metrics at the individual client level [24],”
• Learn from Others and Be Yourself in Heterogeneous Federated Learning (2022, background): “,FedDF [50], FedKT [40] and FEDGEN [75]) have proven the feasibility to do distillation on unlabeled public data or synthetic data.”
• Heterogeneous Federated Learning: State-of-the-art and Research Challenges (2023, methodology): “Unlike these approaches, FedGEN [255] does not rely on the server side to possess a proxy dataset.”
• FedALA: Adaptive Local Aggregation for Personalized Federated Learning (2022, background): “The first one is the pathological heterogeneous setting (McMahan et al. 2017; Shamsian et al. 2021), where we sample 2/2/10 classes for MNIST/Ci-far10/Cifar100 from a total of 10/10/100 classes for each client, with disjoint data samples.”
• Fine-tuning Global Model via Data-Free Knowledge Distillation for Non-IID Federated Learning (2022, methodology): “Recently, FedGen [43] learns a lightweight generator to generate pseudo feature and broadcasts it to clients to regulate local training.”