Time series forecasting is an important and forefront task in many real-world applications. However, most of time series forecasting techniques assume that the training data is clean without anomalies. This assumption is unrealistic since the collected time series data can be contaminated in practice. The forecasting model will be inferior if it is directly trained by time series with anomalies. Thus it is essential to develop methods to automatically learn a robust forecasting model from the contaminated data. In this paper, we first statistically define three types of anomalies, then theoretically and experimentally analyze the loss robustness and sample robustness when these anomalies exist. Based on our analyses, we propose a simple and efficient algorithm to learn a robust forecasting model. Extensive experiments show that our method is highly robust and outperforms all existing approaches. The code is available at https://github.com/haochenglouis/RobustTSF.
When Will It Fail?: Anomaly to Prompt for Forecasting Future Anomalies in Time Series
Department of Artificial Intelligence, Kyung Hee University, Yongin, Republic of Korea(韩国成均馆大学人工智能系);Department of Artificial Intelligence, Korea University, Seoul, Republic of Korea(韩国大学人工智能系)
Monte Carlo EM for Deep Time Series Anomaly Detection
AWS AI Labs(AWS 人工智能实验室);Technical University of Munich(慕尼黑工业大学)
时间序列数据经常被离群值或其他类型的异常所破坏。识别异常点本身可以是一个目标(异常检测),也可以是提高其他时间序列任务(例如预测)性能的一种手段。最近基于深度学习的异常检测和预测方法通常假设训练数据中的异常比例小到足以忽略,并将未标记的数据视为来自标称数据分布。我们提出了一个简单而有效的技术,用于增强现有的时间序列模型,使他们明确占异常的训练数据。通过增加训练数据与潜在的异常指标变量,其分布推断,同时使用Monte Carlo EM训练底层模型,我们的方法同时推断异常点,同时提高模型性能的标称数据。我们证明了该方法的有效性,结合它与一个简单的前馈预测模型。我们调查如何异常的训练集影响训练的预测模型,这是常用的时间序列异常检测,并表明我们的方法提高了训练的模型。
A Deep Learning Model for Forecasting Global Monthly Mean Sea Surface Temperature Anomalies
CSIRO Data61(澳大利亚联邦科学与工业研究组织Data61);The Australian National University(澳大利亚国立大学);Defence Science and Technology Group, Department of Defence(国防部国防科学与技术集团);CSIRO(澳大利亚联邦科学与工业研究组织)
海表温度(SST)变率在全球天气和气候系统中起着关键作用,厄尔尼诺-南方涛动等现象被认为是全球尺度年际气候变率的主要来源。能够对海面温度异常,特别是与极端海洋热浪事件有关的海面温度异常进行长期预报,具有潜在的重大经济和社会效益。我们基于70多年(1950-2021)ECMWF ERA 5月平均平均海面水温和2米气温数据开发了一个深度学习时间序列预测模型(Unet-LSTM)。Unet-LSTM模型能够学习驱动二维全球海面温度时间演变的基础物理学。该模型准确地预测海面温度超过24个月的时间内,均方根误差保持在0.75 $^\circ$ C以下的所有预测的月份。我们还研究了该模型预测Niño3.4区域海表温度异常的能力,以及过去十年中的一些海洋热浪热点。对Niño3.4指数的模型预测使我们能够提前24个月捕捉到2010-11年强烈的拉尼娜现象,2009-10年厄尔尼诺现象和2015-16年极端厄尔尼诺现象。它还显示了对东北太平洋海洋热浪Blob的长期预测技能。然而,对东南部印度洋的海洋热浪--宁格鲁厄尔尼诺现象--的预测显示出有限的技能。这些结果表明,显着的潜力,数据驱动的方法,以产生长期的海表温度异常的预测。
Forecasting Anomaly Precursors via Uncertainty-Aware Time-Series Ensembles
Department of Industrial & Management Engineering, Korea University(韩国大学工业与管理工程系);Department of Industrial Engineering, Seoul National University(首尔国立大学工业工程系);LG CNS
Faculty of Engineering University of Porto(波尔图大学工程学院);Laboratory for Artificial Intelligence and Computer Science (LIACC)(人工智能与计算机科学实验室(LIACC));Fraunhofer Portugal AICOS(弗劳恩霍夫葡萄牙AICOS机构)
CTIA-ODE: Continuous-Time Inter-Variable Attention for Long-Term Time Series Forecasting
UGOD, HKUST-GZ(香港科技大学(广州));School of Software, Yunnan University(云南大学软件学院);UGOD, The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州))
Generating Fine-Grained Causality in Climate Time Series Data for Forecasting and Anomaly Detection
University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);IBM Research(IBM研究院);MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室);Amazon Sustainability Science and Innovation(亚马逊可持续性科学与创新部门)
Forecasting with Deep Learning: Beyond Average of Average of Average Performance
Faculdade de Engenharia da Universidade do Porto(波尔图大学工程学院);Laboratory for Artificial Intelligence and Computer Science (LIACC)(人工智能与计算机科学实验室(LIACC));Fraunhofer Portugal AICOS(弗劳恩霍夫葡萄牙AICOS)
Selective Learning for Deep Time Series Forecasting
State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所人工智能安全国家重点实验室);University of Chinese Academy of Sciences(中国科学院大学);Department of Automation, Tsinghua University(清华大学自动化系)
NetBurst: Event-Centric Forecasting of Bursty, Intermittent Time Series
University of California, Santa Barbara(加州大学圣巴bara分校);Cisco Research(思科研究);Lawrence Berkeley National Laboratory(伯克利国家实验室);Northwestern University(西北大学)
RED-F: Reconstruction-Elimination based Dual-stream Contrastive Forecasting for Multivariate Time Series Anomaly Prediction
School of Computer Science (National Pilot Software Engineering School), Beijing University of Posts and Telecommunications(计算机学院(国家级试点软件工程学院),北京邮电大学);China Telecom Research Institute Beijing(中国电信研究院北京分院);Department of Computer Science, Missouri University of Science and Technology(计算机学院,密苏里科技大学)
DoFlow: Flow-based Generative Models for Interventional and Counterfactual Forecasting on Time Series
H. Milton Stewart School of Industrial and Systems Engineering(H. Milton Stewart工业与系统工程学院);Georgia Institute of Technology(佐治亚理工学院);Northwestern–Argonne Institute for Scientific and Engineering Excellence(西北-阿贡科学与工程卓越研究所)
MamNet: A Novel Hybrid Model for Time-Series Forecasting and Frequency Pattern Analysis in Network Traffic
Columbia University(哥伦比亚大学);University of California, Irvine(加利福尼亚大学欧文分校);Washington University in St. Louis(圣路易斯华盛顿大学);University of Southern California(南加利福尼亚大学);Meta Platform Inc.(Meta平台公司);Northeastern University(东北大学)
Explainable Earth Surface Forecasting under Extreme Events
Image Processing Laboratory (IPL), Universitat de València(瓦伦西亚大学图像处理实验室);Remote Sensing Centre for Earth System Research (RSC4Earth), Leipzig University(莱比锡大学地球系统研究遥感中心);Institute for Earth System Science and Remote Sensing, Leipzig University(莱比锡大学地球系统科学与遥感研究所);Image and Signal Processing Group, Leipzig University(莱比锡大学图像与信号处理组);Helmholtz Centre for Environmental Research – UFZ(亥姆霍兹环境研究中心)
Multi-modal learning for geospatial vegetation forecasting
Max-Planck-Institute for Biogeochemistry(马克斯·普朗克生物地球化学研究所);ELLIS Unit Jena(耶拿ELLIS单元);ETH Zürich(苏黎世联邦理工学院);Hong Kong University of Science and Technology(香港科技大学)
Integrating Artificial Intelligence and Geophysical Insights for Earthquake Forecasting: A Cross-Disciplinary Review
University of Science and Technology Beijing(北京科技大学);Beijing Engineering Research Center of Industrial Spectrum Imaging(北京市工业光谱成像工程研究中心);New York University Abu Dhabi(纽约大学阿布扎比分校);Southern University of Science and Technology(南方科技大学);China University of Geosciences (Beijing)(中国地质大学(北京))