İşbirlikçi Yapay Zeka Konsepti: Federe Öğrenmeye Genel Bir Bakış

Öz

Yapay zeka (YZ) gücünü büyük veriden almaktadır. Ancak büyük veriye ulaşmak ve bu veriyi işlemek, gerek gizlilik, gerekse büyük verinin işlenmesi için gereken donanımsal ihtiyaçlardan ötürü her zaman mümkün olamayabilmektedir. Federe öğrenme (FÖ); bahsi geçen gizlilik & büyük veri ikilemini çözebilmek adına önerilen yeni bir konsepttir. FÖ, ortak bir YZ model parametrelerinin katılımcılar üzerinde güncellenmesi ve güncellenen parametrelerin koordinatör vasıtasıyla birleştirilmesini gerçekleştiren, bunu yaparken de veri gizliliğini koruyan bir çerçevedir. FÖ, mimarisi gereği veri gizliliği korunurken aynı zamanda iş yükü de paylaştırılmış olur. Ayrıca katılımcı sayısı açısından ölçeklenebilirlik ile beraber kimi problemlerde daha yüksek başarım oranı, daha düşük çalışma süreleri gibi avantajlar da sunar. İşbirliği yapan katılımcıların öznitelik ve örnek uzaylarının ne ölçüde ortak olduğuna bağlı olarak yatay, dikey ve transfer FÖ yaklaşımları mevcuttur. Makine öğenmesi yöntemlerinin kullanıldığı ve veri gizliliğinin önem arz ettiği her alanda FÖ kullanım alanı bulmaktadır. Sağlık hizmetleri, nakliye sektörü, finansal teknolojiler ve doğal dil işleme alanları yatay FÖ konseptinin kullanıldığı alanların başında gelmektedir. Öte yandan, dikey ve transfer FÖ konseptleriyle sektörler arasında YZ bazlı işbirlikleri geliştirilebilmektedir.

Anahtar Kelimeler

• Federe Öğrenme
• Merkezi Öğrenme
• Dağıtık Öğrenme
• Büyük Veri
• Veri gizliliği
• Makine Öğrenmesi

Collaborative Artifical Intelligence Concept: Federated Learning Review

Abstract

Artificial intelligence (AI) draws its power from big data. However, accessing and processing big data may not always be possible due to both confidentiality and hardware requirements for high computational performance. Federated learning (FL) is a new concept proposed to solve the aforementioned privacy & big data dilemma. FL is also a framework that performs updating of the parameters of a common AI model trained by the different participants and then combining the updated parameters through the coordinator while protecting data privacy. Due to the modular design of the FL concept, the workload is shared among the participants while protecting data privacy. It also provides advantages like scalability in terms of collaborator count and higher performance and lower execution time for some sort of problems. Depending on the similarity of the feature and sample spaces of the collaborators, there are some FL approaches such as horizontal, vertical and transfer. FL is applicable to any field in which machine learning methods are utilized and the data privacy is an important issue. Healthcare services, transportation sector, financial technologies and natural language processing are the prominent fields where horizontal FL concept is applied. On the other hand, AI-based collaborations between the sectors can be developed with vertical and transfer FL concepts.

Keywords

• Federated Learning
• Centralized Learning
• Distributed Learning
• Big Data
• Data Privacy
• Machine Learning

References

1. [1] E. Hodo, X. Bellekens, A. Hamilton, C. Tachtatzis, and R. Atkinson, “Shallow and Deep Networks Intrusion Detection System: A Taxonomy and Survey,” pp. 1–43, 2017, [Online]. Available: http://arxiv.org/abs/1701.02145.
2. [2] Y. Ma, Z. Wang, H. Yang, and L. Yang, “Artificial intelligence applications in the development of autonomous vehicles: A survey,” IEEE/CAA J. Autom. Sin., vol. 7, no. 2, pp. 315–329, 2020, doi: 10.1109/JAS.2020.1003021.
3. [3] J. Bullock, A. Luccioni, K. H. Pham, C. S. N. Lam, and M. Luengo-Oroz, “Mapping the landscape of artificial intelligence applications against COVID-19,” J. Artif. Intell. Res., vol. 69, pp. 807–845, 2020, doi: 10.1613/JAIR.1.12162.
4. [4] O. Zawacki-Richter, V. I. Marín, M. Bond, and F. Gouverneur, “Systematic review of research on artificial intelligence applications in higher education-where are the educators?,” doi: 10.1186/s41239-019-0171-0.
5. [5] J. Park et al., “Communication-Efficient and Distributed Learning over Wireless Networks: Principles and Applications,” Proc. IEEE, vol. 109, no. 5, pp. 796–819, 2021, doi: 10.1109/JPROC.2021.3055679.
6. [6] “I (Legislative acts) REGULATIONS REGULATION (EU) 2016/679 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (Text with EEA relevance).”
7. [7] H. Brendan McMahan, E. Moore, D. Ramage, S. Hampson, and B. Agüera y Arcas, “Communication-efficient learning of deep networks from decentralized data,” Proc. 20th Int. Conf. Artif. Intell. Stat. AISTATS 2017, vol. 54, 2017.
8. [8] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” Commun. ACM, vol. 60, no. 6, pp. 84–90, Jun. 2017, doi: 10.1145/3065386.

9. [9] Z. Tang, S. Shi, X. Chu, W. Wang, and B. Li, “Communication-Efficient Distributed Deep Learning: A Comprehensive Survey,” no. 1, pp. 1–23, 2020, [Online]. Available: http://arxiv.org/abs/2003.06307.
10. [10] E. M. Campos et al., “Evaluating Federated Learning for intrusion detection in Internet of Things: Review and challenges,” Comput. Networks, vol. 203, p. 108661, Feb. 2022, doi: 10.1016/J.COMNET.2021.108661.
11. [11] P. Kairouz et al., “Advances and open problems in federated learning,” arXiv, pp. 1–105, 2019.
12. [12] D. Jatain, V. Singh, and N. Dahiya, “A contemplative perspective on federated machine learning: Taxonomy, threats & vulnerability assessment and challenges,” J. King Saud Univ. - Comput. Inf. Sci., no. xxxx, 2021, doi: 10.1016/j.jksuci.2021.05.016.
13. [13] Q. Li et al., “A Survey on Federated Learning Systems: Vision, Hype and Reality for Data Privacy and Protection,” IEEE Trans. Knowl. Data Eng., pp. 1–44, 2021, doi: 10.1109/TKDE.2021.3124599.
14. [14] S. Wang et al., “Adaptive Federated Learning in Resource Constrained Edge Computing Systems,” IEEE J. Sel. Areas Commun., vol. 37, no. 6, pp. 1205–1221, 2019, doi: 10.1109/JSAC.2019.2904348.
15. [15] V. Smith, C. Chiang, M. Sanjabi, and A. Talwalkar, “Federated Multi-Task Learning,” no. Nips, 2017.
16. [16] Q. Li, Z. Wen, and B. He, “Practical federated gradient boosting decision trees,” AAAI 2020 - 34th AAAI Conf. Artif. Intell., pp. 4642–4649, 2020, doi: 10.1609/aaai.v34i04.5895.
17. [17] L. Li, Y. Fan, M. Tse, and K. Y. Lin, “A review of applications in federated learning,” Comput. Ind. Eng., vol. 149, no. September, 2020, doi: 10.1016/j.cie.2020.106854.
18. [18] M. Ghassemi, T. Naumann, P. Schulam, A. L. Beam, I. Y. Chen, and R. Ranganath, “A Review of Challenges and Opportunities in Machine Learning for Health.,” AMIA Jt. Summits Transl. Sci. proceedings. AMIA Jt. Summits Transl. Sci., vol. 2020, pp. 191–200, 2020, [Online]. Available: http://www.ncbi.nlm.nih.gov/pubmed/32477638%0Ahttp://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=PMC7233077.
19. [19] Q. Dou et al., “ARTICLE Federated deep learning for detecting COVID-19 lung abnormalities in CT: a privacy-preserving multinational validation study,” doi: 10.1038/s41746-021-00431-6.
20. [20] X. Li, Y. Gu, N. Dvornek, L. H. Staib, P. Ventola, and J. S. Duncan, “Multi-site fMRI analysis using privacy-preserving federated learning and domain adaptation: ABIDE results,” Med. Image Anal., vol. 65, 2020, doi: 10.1016/j.media.2020.101765.
21. [21] L. Huang, A. L. Shea, H. Qian, A. Masurkar, H. Deng, and D. Liu, “Patient clustering improves efficiency of federated machine learning to predict mortality and hospital stay time using distributed electronic medical records,” J. Biomed. Inform., vol. 99, no. September, p. 103291, 2019, doi: 10.1016/j.jbi.2019.103291.
22. [22] K. Tan, D. Bremner, J. Le Kernec, and M. Imran, “Federated Machine Learning in Vehicular Networks: A summary of Recent Applications,” 2020 Int. Conf. UK-China Emerg. Technol. UCET 2020, no. August, 2020, doi: 10.1109/UCET51115.2020.9205482.
23. [23]Y. Liu, J. J. Q. Yu, J. Kang, D. Niyato, and S. Zhang, “Privacy-Preserving Traffic Flow Prediction: A Federated Learning Approach,” IEEE Internet Things J., vol. 7, no. 8, pp. 7751–7763, 2020, doi: 10.1109/JIOT.2020.2991401.
24. [24]A. Nguyen et al., “Deep Federated Learning for Autonomous Driving.” [Online]. Available: https://github.com/aioz-ai/FADNet.
25. [25] A. M. Elbir, B. Soner, and S. Coleri, “Federated Learning in Vehicular Networks,” pp. 1–6, 2020, [Online]. Available: http://arxiv.org/abs/2006.01412.
26. [26] G. Long, T. Shen, Y. Tan, L. Gerrard, A. Clarke, and J. Jiang, “Federated Learning for Privacy-Preserving Open Innovation Future on Digital Health,” Humanit. Driven AI, pp. 113–133, 2022, doi: 10.1007/978-3-030-72188-6_6.
27. [27] A. Imteaj and M. H. Amini, “Leveraging asynchronous federated learning to predict customers financial distress,” Intell. Syst. with Appl., vol. 14, 2022, doi: 10.1016/j.iswa.2022.200064.
28. [28]G. Long, “Federated Learning for Open Banking.”
29. [29]D. G. Bernal, “Decentralizing Large-Scale Natural Language Processing with Federated Learning,” Degree Proj. Comput. Sci. Eng., 2020, [Online]. Available: https://www.diva-portal.org/smash/record.jsf?pid=diva2:1455825.