Mitigating Model Drift in Continuous Machine Learning Pipelines Deployed on Cloud-Native Architectures
Keywords:
Model Drift, MLOps, Cloud-Native Architecture, Continuous Learning, Kubernetes, Data Drift Detection, Concept Drift, Retraining Automation, CI/CD, DevOps for MLAbstract
As machine learning (ML) systems increasingly shift from experimental prototypes to continuously operating services, managing model drift becomes a crucial concern, especially in cloud-native architectures. Model drift, the phenomenon where a model's predictive performance degrades over time due to changes in data distribution, poses significant risks to decision-making systems. This paper explores strategies to mitigate model drift in continuous machine learning pipelines, particularly those deployed on modern cloud-native stacks, including Kubernetes and serverless environments. We propose an integrated framework that leverages drift detection algorithms, retraining triggers, and MLOps automation. The paper also evaluates existing approaches through performance benchmarks and proposes architectural modifications to enhance long-term model reliability.
References
1. Gama, João, et al. “A Survey on Concept Drift Adaptation.” ACM Computing Surveys, vol. 46, no. 4, ACM, pp. 1–37.
2. Kifer, Daniel, Shai Ben-David, and Johannes Gehrke. “Detecting Change in Data Streams.” Proceedings of the Thirtieth International Conference on Very Large Data Bases, VLDB Endowment, pp. 180–191.
3. Sculley, D., et al. “Hidden Technical Debt in Machine Learning Systems.” Advances in Neural Information Processing Systems, vol. 28, Neural Information Processing Systems Foundation, pp. 2503–2511.
4. Žliobaitė, Indrė. “Learning under Concept Drift: An Overview.” arXiv Preprint, arXiv:1010.4784.
5. Baier, Florian, Stefan Reiß, and Sebastian Otte. “Unsupervised Drift Detection with Black-Box Models.” arXiv Preprint, arXiv:2001.08942.
6. Kamadi, S. (2023). Identity-Driven Zero Trust Automation in GitOps: Policy-as-Code Enforcement for Secure Code Deployments. International Journal of Scientific Research in Computer Science, Engineering and Information Technology, 9(3), 893-902. https://doi.org/10.32628/CSEIT235148
7. Lu, Jie, et al. “Learning under Concept Drift: A Review.” IEEE Transactions on Knowledge and Data Engineering, vol. 31, no. 12, IEEE, pp. 2346–2363.
8. Bifet, Albert, and Ricard Gavaldà. “Learning from Time-Changing Data with Adaptive Windowing.” Proceedings of the 2007 SIAM International Conference on Data Mining, SIAM, pp. 443–448.
9. Webb, Geoffrey I., et al. “Characterizing Concept Drift.” Data Mining and Knowledge Discovery, vol. 30, no. 4, Springer, pp. 964–994.
10. Breck, E., et al. “The ML Test Score: A Rubric for ML Production Readiness and Technical Debt Reduction.” Proceedings of the IEEE International Conference on Big Data, IEEE, pp. 1123–1132.
11. Polyzotis, Neoklis, et al. “Data Management Challenges in Production Machine Learning.” Proceedings of the 2017 ACM International Conference on Management of Data, ACM, pp. 1723–1726.
12. Schelter, Sebastian, et al. “Automatically Tracking Metadata and Provenance of Machine Learning Experiments.” Proceedings of the NeurIPS Workshop on ML Systems, Neural Information Processing Systems Foundation.
13. Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin. “Why Should I Trust You? Explaining the Predictions of Any Classifier.” Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ACM, pp. 1135–1144.
Downloads
Published
Issue
Section
License
Copyright (c) 2024 Bertolt Friedrich (Author)
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
