Cloud Resource Auto-Scaling Strategy Based on CNN-Lightweight Transformer

Yue Zhang, Chunhe Song
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:30-35, 2025.

Abstract

With the rapid development of cloud computing and containerization technologies, load forecasting has become increasingly important in resource management. This paper proposes a load forecasting model based on a lightweight Transformer and local convolution fusion, aiming to efficiently capture multi-scale features of complex loads while maintaining low computational overhead. Furthermore, this paper introduces a predictive error feedback and adaptive cooling period adjustment mechanism based on traditional Horizontal Pod Autoscaling (HPA), enhancing the system’s adaptability to load variations by dynamically adjusting scaling strategies. Experimental results demonstrate that the proposed model excels in both load forecasting accuracy and scheduling stability, effectively balancing response speed and system robustness, providing an efficient solution for cloud resource management.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-zhang25a, title = {Cloud Resource Auto-Scaling Strategy Based on CNN-Lightweight Transformer}, author = {Zhang, Yue and Song, Chunhe}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {30--35}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/zhang25a/zhang25a.pdf}, url = {https://proceedings.mlr.press/v278/zhang25a.html}, abstract = { With the rapid development of cloud computing and containerization technologies, load forecasting has become increasingly important in resource management. This paper proposes a load forecasting model based on a lightweight Transformer and local convolution fusion, aiming to efficiently capture multi-scale features of complex loads while maintaining low computational overhead. Furthermore, this paper introduces a predictive error feedback and adaptive cooling period adjustment mechanism based on traditional Horizontal Pod Autoscaling (HPA), enhancing the system’s adaptability to load variations by dynamically adjusting scaling strategies. Experimental results demonstrate that the proposed model excels in both load forecasting accuracy and scheduling stability, effectively balancing response speed and system robustness, providing an efficient solution for cloud resource management.} }
Endnote
%0 Conference Paper %T Cloud Resource Auto-Scaling Strategy Based on CNN-Lightweight Transformer %A Yue Zhang %A Chunhe Song %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-zhang25a %I PMLR %P 30--35 %U https://proceedings.mlr.press/v278/zhang25a.html %V 278 %X With the rapid development of cloud computing and containerization technologies, load forecasting has become increasingly important in resource management. This paper proposes a load forecasting model based on a lightweight Transformer and local convolution fusion, aiming to efficiently capture multi-scale features of complex loads while maintaining low computational overhead. Furthermore, this paper introduces a predictive error feedback and adaptive cooling period adjustment mechanism based on traditional Horizontal Pod Autoscaling (HPA), enhancing the system’s adaptability to load variations by dynamically adjusting scaling strategies. Experimental results demonstrate that the proposed model excels in both load forecasting accuracy and scheduling stability, effectively balancing response speed and system robustness, providing an efficient solution for cloud resource management.
APA
Zhang, Y. & Song, C.. (2025). Cloud Resource Auto-Scaling Strategy Based on CNN-Lightweight Transformer. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:30-35 Available from https://proceedings.mlr.press/v278/zhang25a.html.

Related Material