arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2205.05511cs.LG
收藏

用于时间序列预测的高效自动化深度学习

Efficient Automated Deep Learning for Time Series Forecasting

Leibniz University Hannover(汉诺威莱布尼茨大学);Ludwig-Maximilian University(路德维希-马克西米利安大学);Fraunhofer Institute for Integrated Circuits (IIS)(弗劳恩霍夫集成电路研究所);University of Freiburg(弗莱堡大学);Bosch Center for Artificial Intelligence(博世人工智能中心)

Difan Deng, Florian Karl, Frank Hutter, Bernd Bischl, Marius Lindauer

摘要:

近年来,自动机器学习(AutoML)的效率得到了极大的提高,特别是自动深度学习(AutoDL)系统,但最近的工作集中在表格,图像或NLP任务上。到目前为止,很少有人注意到一般的AutoDL框架的时间序列预测,尽管在应用不同的新架构,这些任务的巨大成功。在本文中,我们提出了一种有效的方法,用于联合优化神经结构和超参数的整个数据处理管道的时间序列预测。与常见的NAS搜索空间相比,我们设计了一种新的神经架构搜索空间,覆盖各种最先进的架构,允许在不同的DL方法上进行有效的宏搜索。为了有效地搜索在这样一个大的配置空间,我们使用贝叶斯优化与多保真度优化。我们实证研究了几种不同的预算类型,使有效的多保真度优化不同的预测数据集。此外,我们将我们的系统(称为\system)与几个已建立的基线进行了比较,并表明它在几个数据集上的性能明显优于所有基线。

英文摘要:

Recent years have witnessed tremendously improved efficiency of Automated Machine Learning (AutoML), especially Automated Deep Learning (AutoDL) systems, but recent work focuses on tabular, image, or NLP tasks. So far, little attention has been paid to general AutoDL frameworks for time series forecasting, despite the enormous success in applying different novel architectures to such tasks. In this paper, we propose an efficient approach for the joint optimization of neural architecture and hyperparameters of the entire data processing pipeline for time series forecasting. In contrast to common NAS search spaces, we designed a novel neural architecture search space covering various state-of-the-art architectures, allowing for an efficient macro-search over different DL approaches. To efficiently search in such a large configuration space, we use Bayesian optimization with multi-fidelity optimization. We empirically study several different budget types enabling efficient multi-fidelity optimization on different forecasting datasets. Furthermore, we compared our resulting system, dubbed \system, against several established baselines and show that it significantly outperforms all of them across several datasets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑