Bayesian Stream Tuner: Dynamic Hyperparameter Optimization for Real-Time Data Streams
Uloženo v:
| Název: | Bayesian Stream Tuner: Dynamic Hyperparameter Optimization for Real-Time Data Streams |
|---|---|
| Autoři: | Nilesh Verma, Albert Bifet, Bernhard Pfahringer, Maroua Bahri |
| Přispěvatelé: | Bahri, Maroua |
| Zdroj: | Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2. :2871-2882 |
| Informace o vydavateli: | ACM, 2025. |
| Rok vydání: | 2025 |
| Témata: | [INFO.INFO-AI] Computer Science [cs]/Artificial Intelligence [cs.AI], Bayesian Optimization, Real-time Learning, Online Hyperparameter Optimization, [INFO.INFO-LG] Computer Science [cs]/Machine Learning [cs.LG], Adaptive Learning, Data Stream Mining |
| Popis: | Hyperparameter optimization is crucial for maximizing machine learning model performance, yet most existing algorithms are designed for batch or offline scenarios and assume static data distributions. Such assumptions fall short in data stream settings, where models must adapt to evolving inputs in real time. To address these limitations, we propose the Bayesian Stream Tuner (BST), a novel framework for online hyperparameter optimization in nonstationary data streams. BST maintains a dynamic set of candidate hyperparameter configurations and periodically refines them using an incremental Bayesian model, which estimates configuration performance based on recent data statistics and hyperparameter values. This systematic exploration and refinement strategy allows BST to detect and respond to concept drift by resetting its adaptation mechanisms whenever necessary, ensuring strong performance under changing distributions. Our theoretical analysis establishes sublinear regret bounds for BST in dynamic environments, and extensive experiments on classification and regression tasks demonstrate that BST consistently outperforms state-of-the-art online hyperparameter optimization methods in both predictive accuracy and adaptability, making it a powerful solution for real-time hyperparameter tuning in evolving data streams. |
| Druh dokumentu: | Article Conference object |
| Popis souboru: | application/pdf |
| DOI: | 10.1145/3711896.3736852 |
| Přístupová URL adresa: | https://hal.science/hal-05210830v1 https://hal.science/hal-05210830v1/document https://doi.org/10.1145/3711896.3736852 |
| Přístupové číslo: | edsair.doi.dedup.....6cb18dcbfec7853f82ac8441115ed70c |
| Databáze: | OpenAIRE |
Nájsť tento článok vo Web of Science | Abstrakt: | Hyperparameter optimization is crucial for maximizing machine learning model performance, yet most existing algorithms are designed for batch or offline scenarios and assume static data distributions. Such assumptions fall short in data stream settings, where models must adapt to evolving inputs in real time. To address these limitations, we propose the Bayesian Stream Tuner (BST), a novel framework for online hyperparameter optimization in nonstationary data streams. BST maintains a dynamic set of candidate hyperparameter configurations and periodically refines them using an incremental Bayesian model, which estimates configuration performance based on recent data statistics and hyperparameter values. This systematic exploration and refinement strategy allows BST to detect and respond to concept drift by resetting its adaptation mechanisms whenever necessary, ensuring strong performance under changing distributions. Our theoretical analysis establishes sublinear regret bounds for BST in dynamic environments, and extensive experiments on classification and regression tasks demonstrate that BST consistently outperforms state-of-the-art online hyperparameter optimization methods in both predictive accuracy and adaptability, making it a powerful solution for real-time hyperparameter tuning in evolving data streams. |
|---|---|
| DOI: | 10.1145/3711896.3736852 |