响马读paper

一个要求成员每月至少读一篇文献并打卡的学术交流社群

2020, International Journal of Forecasting. DOI: 10.1016/j.ijforecast.2019.07.001
DeepAR: Probabilistic forecasting with autoregressive recurrent networks
David Salinas, Valentin Flunkert, Jan Gasthaus, Tim Januschowski
Abstract:
Probabilistic forecasting, i.e., estimating a time series’ future probability distribution given its past, is a key enabler for optimizing business processes. In retail businesses, for example, probabilistic demand forecasts are crucial for having the right inventory available at the right time and in the right place. This paper proposes DeepAR, a methodology for producing accurate probabilistic forecasts, based on training an autoregressive recurrent neural network model on a large number of related time series. We demonstrate how the application of deep learning techniques to forecasting can overcome many of the challenges that are faced by widely-used classical approaches to the problem. By means of extensive empirical evaluations on several real-world forecasting datasets, we show that our methodology produces more accurate forecasts than other state-of-the-art methods, while requiring minimal manual work.
2022-04-23 15:36:00
#paper Salinas, D., Flunkert, V., Gasthaus, J., & Januschowski, T. (2020). DeepAR: Probabilistic forecasting with autoregressive recurrent networks. International Journal of Forecasting, 36(3), 1181–1191. https://doi.org/10.1016/j.ijforecast.2019.07.001 2020的DeepAR算法,个人觉得还是蛮有启发性的。相比于大部分在时序问题这种的点预测,DeepAR用概率模型的思路,在每个时间点去预测期概率分布;这其实也可能更符合显示,毕竟本身时序过程就是有非常强的随机属性,概率分布本身也更贴近本质。文章本身对鲁棒性讨论不多,但DeepAR的鲁棒性应该比较好。另外就是DeepAR自身强调的是,他可以很方便地对多个相关的序列(数千上万)个进行建模并提取其中的关系,这一点确实也是比较强的。所以作者也特别提到,仅需要少量的特征工程及超参数调整,即能获得比传统模型更好的效果。(论文中的模型对比,我个人觉得确实也相对规范)。 论文本身写得很精炼,但是因为是Amazon的论文,所以亲生儿子是用Mxnet上搭建的,用起来确实有点不太方便。Pytprch和TF倒是有实现,但是实现细节也有些魔改的地方。方便性来看,确实比不过Prophet,哈哈
TOP