Journal of Internet Computing and Services (인터넷정보학회논문지)

Korean Society for Internet Information (한국인터넷정보학회)
권호 표지
DOI QR코드

DOI QR Code

Federated learning-based client training acceleration method for personalized digital twins

개인화 디지털 트윈을 위한 연합학습 기반 클라이언트 훈련 가속 방식

  • YoungHwan Jeong (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI)) ;
  • Won-gi Choi (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI)) ;
  • Hyoseon Kye (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI)) ;
  • JeeHyeong Kim (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI)) ;
  • Min-hwan Song (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI)) ;
  • Sang-shin Lee (Autonomous Intelligence System Research Center, Korea Electronics Technology Institute (KETI))
  • 정영환 ;
  • 최원기 ;
  • 계효선 ;
  • 김지형 ;
  • 송민환 ;
  • 이상신
  • Received : 2024.03.21
  • Accepted : 2024.07.18
  • Published : 2024.08.31
Download PDF
⟨ Previous Next ⟩

Abstract

Digital twin is an M&S (Modeling and Simulation) technology designed to solve or optimize problems in the real world by replicating physical objects in the real world as virtual objects in the digital world and predicting phenomena that may occur in the future through simulation. Digital twins have been elaborately designed and utilized based on data collected to achieve specific purposes in large-scale environments such as cities and industrial facilities. In order to apply this digital twin technology to real life and expand it into user-customized service technology, practical but sensitive issues such as personal information protection and personalization of simulations must be resolved. To solve this problem, this paper proposes a federated learning-based accelerated client training method (FACTS) for personalized digital twins. The basic approach is to use a cluster-driven federated learning training procedure to protect personal information while simultaneously selecting a training model similar to the user and training it adaptively. As a result of experiments under various statistically heterogeneous conditions, FACTS was found to be superior to the existing FL method in terms of training speed and resource efficiency.

디지털 트윈은 현실세계의 물리적 객체를 디지털 세계의 가상객체로 모사하고 시뮬레이션을 통해 미래에 발생 가능한 현상을 예측함으로써, 현실세계의 문제를 해결 또는 최적화하기 위해 고안된 M&S(Modeling and Simulation) 기술이다. 디지털 트윈은 지금까지 도시, 산업 시설 등 대규모 환경에서 특정 목적을 달성하기 위해 수집된 다양한 데이터 기반으로 정교하게 설계되고 활용되어 왔다. 이러한 디지털 트윈 기술을 실생활에 적용하고 사용자 맞춤형 서비스 기술로 확장하기 위해서는 개인정보 보호, 시뮬레이션의 개인화 등 실질적이지만 민감한 문제를 해결해야 한다. 이러한 문제를 해결하기 위해 본 논문에서는 개인화 디지털 트윈을 위한 연합학습 기반의 클라이언트 훈련 가속 방식(FACTS)을 제안한다. 기본적인 접근 방식은 클러스터 기반의 적응형 연합학습 훈련 절차를 활용해 개인정보를 보호하면서 동시에 사용자와 유사한 훈련 모델을 선택하고 훈련을 가속하는 것이다. 다양한 통계적으로 이질적인 조건의 실험 결과 FACTS는 기존의 FL 방식에 비해 훈련 속도 및 자원 효율성 측면에서 우수한 것으로 나타난다.

Keywords

Acknowledgement

이 논문은 2024년도 정부(과학기술정보통신부)의 재원으로 정보통신기획평가원의 지원을 받아 수행된 연구임 (No.RS-2022-II220545, 지능형 디지털 트윈 연합 객체 구성 및 데이터 프로세싱 기술 개발)

References

  1. Jones, D., Snider, C., Nassehi, A., Yon, J. and Hicks, B., "Characterising the Digital Twin: A systematic literature review," CIRP Journal of Manufacturing Science and Technology, Vol. 29, pp. 36-52, 2020. https://doi.org/10.1016/j.cirpj.2020.02.002
  2. 정득영 외, "디지털 트윈 기술 K-로드맵 ver 1.0," 정보 통신기획평가원, 2021. https://iitp.kr/kr/1/knowledge/openReference.it
  3. Barricelli, B. R., Casiraghi, E. and Fogli, D., "A survey on digital twin: Definitions, characteristics, applications, and design implications," IEEE Access, Vol. 7, pp. 167653-167671, 2019. https://doi.org/10.1109/ACCESS.2019.2953499
  4. Fang, H., and Qian, Q., "Privacy preserving machine learning with homomorphic encryption and federated learning," Future Internet, Vol. 13, No.4, 2021. https://doi.org/10.3390/fi13040094
  5. McMahan, B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A., "Communication-efficient learning of deep networks from decentralized data," In Artificial intelligence and statistics, PMLR, pp. 1273-1282, 2017. https://proceedings.mlr.press/v54/mcmahan17a?ref=https://githubhelp.com
  6. Vepakomma, P., Gupta, O., Swedish, T., and Raskar, R., "Split learning for health: Distributed deep learning without sharing raw patient data," arXiv preprint arXiv:1812.00564, 2018. https://doi.org/10.48550/arXiv.1812.00564
  7. Thapa, C., Arachchige, P. C. M., Camtepe, S., & Sun, L. "Splitfed: When federated learning meets split learning," In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36, No. 8, pp. 8485-8493, 2022. https://doi.org/10.1609/aaai.v36i8.20825
  8. Grieves, M. "Digital twin: manufacturing excellence through virtual factory replication," White paper, Vol. 1, pp. 1-7, 2014. https://doi.org/10.5281/zenodo.1493930
  9. He, Y., Guo, J. and Zheng, X., "From surveillance to digital twin: Challenges and recent advances of signal processing for industrial internet of things," IEEE Signal Processing Magazine, Vol. 35, No. 5, pp. 120-129, 2018. https://doi.org/10.1109/MSP.2018.2842228
  10. Bruynseels, K., Santoni de Sio, F. and Van den Hoven, J., "Digital twins in health care: ethical implications of an emerging engineering paradigm," Frontiers in genetics, Vol. 9, No. 31, 2018 https://doi.org/10.3389/fgene.2018.00031
  11. Singh, M., Fuenmayor, E., Hinchy, E. P., Qiao, Y., Murray, N., and Devine, D., "Digital twin: Origin to future," Applied System Innovation, Vol. 4, No. 2, 2021. https://doi.org/10.3390/asi4020036
  12. Han, Y., Niyato, D., Leung, C., Kim, D. I., Zhu, K., Feng, S. and Miao, C., "A dynamic hierarchical framework for IoT-assisted digital twin synchronization in the metaverse," IEEE Internet of Things Journal, Vol. 10, No. 1, pp. 268-284, 2022. https://doi.org/10.1109/JIOT.2022.3201082
  13. Jia, P., Wang, X., and Shen, X. "Digital-twin-enabled intelligent distributed clock synchronization in industrial IoT systems," IEEE Internet of Things Journal, Vol. 8, No. 6, pp. 4548-4559, 2020. https://doi.org/10.1109/JIOT.2020.3029131
  14. Jiang, Y., Li, M., Li, M., Liu, X., Zhong, R. Y., Pan, W. and Huang, G. Q., "Digital twin-enabled real-time synchronization for planning, scheduling, and execution in precast on-site assembly," Automation in Construction, Vol. 141, No. 104397, 2022. https://doi.org/10.1016/j.autcon.2022.104397
  15. Elayan, H., Aloqaily, M., and Guizani, M., "Digital twin for intelligent context-aware IoT healthcare systems" IEEE Internet of Things Journal, Vol. 8, No. 23, pp. 16749-16757, 2021. https://doi.org/10.1109/JIOT.2021.3051158
  16. Kharchenko, V., Illiashenko, O., Morozova, O., and Sokolov, S., "Combination of digital twin and artificial intelligence in manufacturing using industrial IoT," In IEEE 11th international conference on dependable systems, services and technologies, pp. 196-201, IEEE, 2020. https://doi.org/10.1109/DESSERT50317.2020.9125038
  17. Zhou, Y., Xing, T., Song, Y., Li, Y., Zhu, X., Li, G. and Ding, S., "Digital-twin-driven geometric optimization of centrifugal impeller with free-form blades for five-axis flank milling," Journal of Manufacturing Systems, Vol. 58, pp. 22-35, 2021. https://doi.org/10.1016/j.jmsy.2020.06.019
  18. Rao, D. J. and Mane, S., "Digital twin approach to clinical dss with explainable ai," arXiv preprint arXiv:1910.13520, 2019. https://arxiv.org/abs/1910.13520 https://doi.org/10.13520
  19. Peter, K. McMahan, B. and Brendan, A. et al., "Advances and open problems in federated learning," Found. Trends Mach. Learn., Vol. 14, pp. 1-210, 2021. http://dx.doi.org/10.1561/2200000083
  20. Jakub, K., McMahan, B., Daniel, R. et al., "Federated optimization: Distributed machine learning for on-device intelligence," arXiv:1610.02527, 2016. https://doi.org/10.48550/arXiv.1610.02527
  21. Nguyen, D.C., Ding, M., Pathirana, P.N., Seneviratne, A., Li, J. and Poor, H.V., "Federated learning for internet of things: A comprehensive survey," IEEE Commun. Surv. Tutor, Vol. 23, pp. 1622-1658, 2021. https://doi.org/10.1109/COMST.2021.3075439
  22. Amirhossein, R., Isidoros, T., Hamed, H., Aryan, M. and Ramtin, P., "Straggler-Resilient Federated Learning: Leveraging the Interplay Between Statistical Accuracy and System Heterogeneity," In Proceedings of the 38th International Conference on Machine Learning, Virtual, pp. 18-24, 2021. https://doi.org/10.1109/JSAIT.2022.3205475
  23. Tao, Y. and Zhou, J., "Straggler Remission for Federated Learning via Decentralized Redundant Cayley Tree," In Proceedings of the 2020 IEEE Latin-American Conference on Communications (LATINCOM), pp. 1-6, 2020. https://doi.org/10.1109/LATINCOM50620.2020.9282334
  24. Li, T., Sahu, A.K., Sanjabi, M., Zaheer, M., Talwalker, A. and Smith, V., "Federated optimization in heterogeneous networks," Proc. Mach. Learn. Syst., Vol. 2, pp. 429-450, 2020. https://doi.org/10.48550/arXiv.1812.06127
  25. Yang, H., Fang, M. and Liu, J., "Achieving Linear Speedup with Partial Worker Participation in Non-IID Federated Learning," In Proceedings of the 9th International Conference on Learning Representations, Virtual, pp. 3-7, 2021. https://doi.org/10.48550/arXiv.2101.11203
  26. Felix, S., Simon, W., Klaus, R.M. and Wojciech, S., "Robust and Communication-Efficient Federated Learning From Non-i.i.d. Data," IEEE Trans. Neural Netw. Learn. Syst. Vol. 31, pp. 3400-3413, 2020. https://doi.org/10.1109/TNNLS.2019.2944481
  27. Karimireddy, S. P., Kale, S., Mohri, M., Reddi, S. J., Stich, S. U., & Suresh, A. T., "SCAFFOLD: Stochastic Controlled Averaging for Federated Learning," arXiv preprint arXiv:1910.06378, 2019. https://doi.org/10.48550/arXiv.1910.06378
  28. Lai, F., Zhu, X., Madhyastha, H.V. and Chowdhury, M., "Oort: Efficient federated learning via guided participant selection," In Proceedings of the 15th USENIX Symposium on Operating Systems Design and Implementation, pp. 19-35, 2021. https://www.usenix.org/conference/osdi21/presentation/lai
  29. Taipalus, Toni, "Vector database management systems: Fundamental concepts, use-cases, and current challenges," Cognitive Systems Research, No. 101216, 2024. https://doi.org/10.1016/j.cogsys.2024.101216
  30. Han, Yikun, Chunjiang Liu, and Pengfei Wang, "A comprehensive survey on vector database: Storage and retrieval technique, challenge," arXiv preprint arXiv: 2310.11703, 2023. https://doi.org/10.48550/arXiv.2310.11703
  31. Taipalus, Toni, "Vector database management systems: Fundamental concepts, use-cases, and current challenges," arXiv preprint arXiv:2309.11322, 2023. https://arxiv.org/abs/2309.11322
  32. Andoni, Alexandr et al., "Practical and optimal LSH for angular distance," in Proc. of Advances in neural information processing systems, Vol. 28, 2015. https://doi.org/10.48550/arXiv.1509.02897
  33. Jegou, Herve, Matthijs Douze, and Cordelia Schmid, "Product quantization for nearest neighbor search," IEEE transactions on pattern analysis and machine intelligence, Vol. 33 No. 1, pp. 117-128, 2011. https://doi.org/10.1109/TPAMI.2010.57
  34. Malkov, Yu A., and Dmitry A. Yashunin, "Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs," IEEE transactions on pattern analysis and machine intelligence, Vol. 42, No. 4, pp. 824-836, 2020. https://doi.org/10.1109/TPAMI.2018.2889473
  35. Milvus. (n.d.). Milvus. Retrieved June 7, 2024, https://milvus.io/
  36. Chroma. (n.d.). Chroma. Retrieved June 7, 2024, https://www.trychroma.com/
  37. Pinecone. (n.d.). Pinecone. Retrieved June 7, 2024, https://www.pinecone.io/
  38. Weaviate. (n.d.). Weaviate. Retrieved June 7, 2024, https://weaviate.io/
  39. Redis Labs. (n.d.). Redis. Retrieved June 7, 2024, https://redis.io/