Federated learning algorithm based on knowledge distillation

Federated learning is a new scheme of distributed machine learning, which enables a large number of edge computing devices to jointly learn a shared model without private data sharing. Federated learning allows nodes to synchronize only the locally trained models instead of their own private data, w...

Full description

Bibliographic Details
Main Authors: Jiang, Donglin, Shan, Chen, Zhang, Zhihui
Format: Conference or Workshop Item
Language:English
Published: 2021
Subjects:
Online Access:https://eprints.nottingham.ac.uk/65181/
_version_ 1848800196146757632
author Jiang, Donglin
Shan, Chen
Zhang, Zhihui
author_facet Jiang, Donglin
Shan, Chen
Zhang, Zhihui
author_sort Jiang, Donglin
building Nottingham Research Data Repository
collection Online Access
description Federated learning is a new scheme of distributed machine learning, which enables a large number of edge computing devices to jointly learn a shared model without private data sharing. Federated learning allows nodes to synchronize only the locally trained models instead of their own private data, which provides a guarantee for privacy and security. However, due to the challenges of heterogeneity in federated learning, which are: (1) heterogeneous model architecture among devices; (2) statistical heterogeneity in real federated dataset, which do not obey independent-identical-distribution, resulting in poor performance of traditional federated learning algorithms. To solve the problems above, this paper proposes FedDistill, a new distributed training method based on knowledge distillation. By introducing personalized model on each device, the personalized model aims to improve the local performance even in a situation that global model fails to adapt to the local dataset, thereby improving the ability and robustness of the global model. The improvement of the performance of local device benefits from the effect of knowledge distillation, which can guide the improvement of global model by knowledge transfer between heterogeneous networks. Experiments show that FedDistill can significantly improve the accuracy of classification tasks and meet the needs of heterogeneous users.
first_indexed 2025-11-14T20:47:43Z
format Conference or Workshop Item
id nottingham-65181
institution University of Nottingham Malaysia Campus
institution_category Local University
language English
last_indexed 2025-11-14T20:47:43Z
publishDate 2021
recordtype eprints
repository_type Digital Repository
spelling nottingham-651812021-05-07T08:51:01Z https://eprints.nottingham.ac.uk/65181/ Federated learning algorithm based on knowledge distillation Jiang, Donglin Shan, Chen Zhang, Zhihui Federated learning is a new scheme of distributed machine learning, which enables a large number of edge computing devices to jointly learn a shared model without private data sharing. Federated learning allows nodes to synchronize only the locally trained models instead of their own private data, which provides a guarantee for privacy and security. However, due to the challenges of heterogeneity in federated learning, which are: (1) heterogeneous model architecture among devices; (2) statistical heterogeneity in real federated dataset, which do not obey independent-identical-distribution, resulting in poor performance of traditional federated learning algorithms. To solve the problems above, this paper proposes FedDistill, a new distributed training method based on knowledge distillation. By introducing personalized model on each device, the personalized model aims to improve the local performance even in a situation that global model fails to adapt to the local dataset, thereby improving the ability and robustness of the global model. The improvement of the performance of local device benefits from the effect of knowledge distillation, which can guide the improvement of global model by knowledge transfer between heterogeneous networks. Experiments show that FedDistill can significantly improve the accuracy of classification tasks and meet the needs of heterogeneous users. 2021-03-01 Conference or Workshop Item PeerReviewed application/pdf en cc_by https://eprints.nottingham.ac.uk/65181/1/Title%20Pages%20Example%20%200.6-%E5%B7%B2%E8%9E%8D%E5%90%88%20%284%29.pdf Jiang, Donglin, Shan, Chen and Zhang, Zhihui (2021) Federated learning algorithm based on knowledge distillation. In: 2020 International Conference on Artificial Intelligence and Computer Engineering (ICAICE). Federated learning; Knowledge distillation; Non-independent-identical-distribution; Heterogeneous network http://dx.doi.org/10.1109/ICAICE51518.2020.00038 10.1109/ICAICE51518.2020.00038 10.1109/ICAICE51518.2020.00038 10.1109/ICAICE51518.2020.00038
spellingShingle Federated learning; Knowledge distillation; Non-independent-identical-distribution; Heterogeneous network
Jiang, Donglin
Shan, Chen
Zhang, Zhihui
Federated learning algorithm based on knowledge distillation
title Federated learning algorithm based on knowledge distillation
title_full Federated learning algorithm based on knowledge distillation
title_fullStr Federated learning algorithm based on knowledge distillation
title_full_unstemmed Federated learning algorithm based on knowledge distillation
title_short Federated learning algorithm based on knowledge distillation
title_sort federated learning algorithm based on knowledge distillation
topic Federated learning; Knowledge distillation; Non-independent-identical-distribution; Heterogeneous network
url https://eprints.nottingham.ac.uk/65181/
https://eprints.nottingham.ac.uk/65181/
https://eprints.nottingham.ac.uk/65181/