代码以图示为准
总结
本文介绍了自动化超参数优化赛道的冠军方案,包括贝叶斯优化算法和早停方法。很幸运能够拿到此次比赛的冠军。感谢赛事主办方为我们提供了富有现实意义的比赛场景,让我们积累了宝贵的比赛经验和超参数优化实际经验。希望我们的分享能够对大家有所帮助。
引用
[1] 黑盒优化系统 OpenBox
https://github.com/PKU-DAIR/open-box
[2] 自动化机器学习系统 MindWare
https://github.com/PKU-DAIR/mindware
[3] 比赛冠军源码
https://github.com/PKU-DAIR/2021_AIAC_Task2_1st
[4] https://github.com/automl/ConfigSpace
[5] Liam Li, Kevin Jamieson, Afshin Rostamizadeh, Ekaterina Gonina, Jonathan Bentzur, Moritz Hardt, Benjamin Recht, and Ameet Talwalkar. 2020. A System for Massively Parallel Hyperparameter Tuning. Proceedings of Machine Learning and Systems 2 (2020), 230–246.
[6] Yang Li, Yu Shen, Jiawei Jiang, Jinyang Gao, Ce Zhang, and Bin Cui. 2021. MFES-HB: Efficient Hyperband with Multi-Fidelity Quality Measurements. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 8491–8500.
[7] Yang Li, Yu Shen, Wentao Zhang, Yuanwei Chen, Huaijun Jiang, Mingchao Liu, Jiawei Jiang, Jinyang Gao, Wentao Wu, Zhi Yang, Ce Zhang, and Bin Cui. 2021. OpenBox: A Generalized Black-box Optimization Service. Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (2021).
[8] Yang Li, Yu Shen, Wentao Zhang, Jiawei Jiang, Bolin Ding, Yaliang Li, Jingren Zhou, Zhi Yang, Wentao Wu, Ce Zhang, and Bin Cui. 2021. VolcanoML: Speeding up End-to-End AutoML via Scalable Search Space Decomposition. Proceedings of VLDB Endowment 14 (2021), 2167–2176.
[9]QQ 浏览器 2021AI 算法大赛
https://algo.browser.qq.com/