Long-term time-series forecasting supports a wide range of applications, including weather prediction and electricity demand planning. Frequency-domain methods address this task by decomposing observations into components that describe temporal variations at different scales. However, separate representations do not by themselves provide an explicit mechanism for adjusting the training emphasis across components. Under a shared forecasting objective, the frequency-specific networks can retain different levels of coefficient prediction error, motivating an error-dependent adjustment to their gradients. To this end, we propose BEAT (Balanced frEquency Adaptive Tuning), a framework that combines frequency-specific error monitoring with adaptive gradient modulation. We design a Frequency-Specific Monitor that compares predicted and target wavelet coefficients in a common normalized space and expresses each discrepancy relative to a reference error computed from the detail components. We further introduce a Dynamical Gradient Balancer that converts these ratios into positive, bounded coefficients. Components with higher relative errors receive larger gradient weights, whereas those with lower relative errors receive smaller weights. A shared modulation-strength parameter controls the departure from unmodulated training, and the monitoring and balancing operations are used only during training. Experiments on seven real-world datasets show that BEAT achieves competitive performance against state-of-the-art forecasting methods.
BLAST: Balanced Sampling Time Series Corpus for Universal Forecasting Models
Institute of Computing Technology, Chinese Academy of Sciences(计算技术研究所,中国科学院);State Key Laboratory of AI Safety(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)
Forecasting Source Stability in Scientific Experiments using Temporal Learning Models: A Case Study from Tritium Monitoring
Institute for Data Processing(数据处理研究所);Schmalkalden University of Applied Sciences(施马尔卡尔登应用科学大学);Technical University Munich(慕尼黑技术大学);Max-Planck-Institut für Kernphysik(马克斯·普朗克核物理研究所)
Comparative Analysis of Modern Machine Learning Models for Retail Sales Forecasting
Faculty of Electrical Engineering and Computing, University of Zagreb(Zagreb大学电气工程与计算学院);It From Bit d.o.o.(It From Bit公司);Babson College(巴布森学院);mStart Plus d.o.o.(mStart Plus公司)
Curse of Attention: A Kernel-Based Perspective for Why Transformers Fail to Generalize on Time Series Forecasting and Beyond
The University of Hong Kong(香港大学);University of Wisconsin-Madison(威斯康星大学麦迪逊分校);The Simons Institute for the Theory of Computing(西蒙斯理论计算研究所);UC Berkeley(加州大学伯克利分校);Sun Yat-sen University(中山大学)
Short-term load forecasting under EU-AI Act Requirements in Safety-Critical Environments: Results from a 41-day live challenge on the aggregated German transmission-grid load
Salt-Rock Creep Deformation Forecasting Using Deep Neural Networks and Analytical Models for Subsurface Energy Storage Applications
IIT (ISM) Dhanbad(印度理工学院(印度矿业学院)丹巴德分校);Sorbonne University, Abu Dhabi(索邦大学阿布扎比校区);Sorbonne Center for Artificial Intelligence, Sorbonne University(索邦大学人工智能中心);CSIRO Energy(澳大利亚联邦科学与工业研究组织能源研究所)
DSAT-HD: Dual-Stream Adaptive Transformer with Hybrid Decomposition for Multivariate Time Series Forecasting
Computer Science and Technology(计算机科学与技术);Harbin Engineering University(哈尔滨工程大学);Traditional Chinese Medicine Data Center(传统中医数据中心);China Academy of Chinese Medical Sciences(中国中医科学院)
N-BEATS-MOE: N-BEATS with a Mixture-of-Experts Layer for Heterogeneous Time Series Forecasting
Faculdade de Engenharia da Universidade do Porto, Porto, Portugal(葡萄牙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)
Macroeconomic forecasting with LSTM and mixed frequency time series data
Faculty of Economics, Khon Kaen University(孔敬大学经济学院);Khon Kaen University(孔敬大学)
本文展示了长短期记忆(LSTM)在应用于不同频率采样的宏观经济时间序列数据时的潜力。我们首先介绍了当对所有低频输出和高频变量对应用相同的失配比时,传统的LSTM模型如何适应在混合频率下观察到的时间序列。为了将LSTM推广到多个失配率的情况,我们采用了无限制混合数据采样(U-MIDAS)方案(Foroni et al.,2015年)引入LSTM架构。我们评估通过bothMonte Carlo模拟和实证应用的样本预测性能。我们提出的模型优于限制MIDAS模型,即使在一个设置有利于MIDAS估计。对于现实世界的应用,我们研究预测泰国实际GDP的季度增长率,使用大量的宏观经济指标季度和月度。我们使用U-MIDAS方案的LSTM在所有范围内都轻松击败了简单的基准AR(1)模型,但仅在一个月和六个月的时间内优于强基准单变量LSTM。尽管如此,我们发现,我们提出的模型可能是非常有帮助的,在大的经济低迷时期的短期预测。模拟和实验结果似乎支持我们提出的LSTM with U-MIDAS方案用于临近预报应用。
Luddy School of Informatics, Computing, and Engineering, Indiana University Bloomington(印第安纳大学伯明顿分校勒迪信息学、计算与工程学院);University of Illinois Springfield(伊利诺伊大学斯普林菲尔德分校);College of Science and Engineering, Hamad Bin Khalifa University(哈马德·本·哈利法大学科学与工程学院)
Learning Temporal Saliency for Time Series Forecasting with Cross-Scale Attention
Department of Software Engineering, Faculty of Computer and Information Sciences, Sakarya University(软件工程系,计算机与信息科学学院,萨卡里亚大学);Department of Computer and Information Science (IDA), REAL, AIICS, Linkoping University(计算机与信息科学系(IDA),REAL,AIICS,林奈大学)
Uncertainty-Aware End-to-End AI Weather Forecasting: Disentangling Observation and Model Contributions
Fraunhofer Heinrich-Hertz Institute(弗劳恩霍夫海因里希·赫兹研究所);Technische Universität Berlin(柏林工业大学);BIFOLD - Berlin Institute for the Foundations of Learning and Data(柏林学习与数据基础研究所(BIFOLD))
Forecasting India's Demographic Transition Under Fertility Policy Scenarios Using hybrid LSTM-PINN Model
Sant Longowal Institute of Engineering and Technology(桑特·朗戈瓦尔工程与技术学院);School of Computer and Mathematical Sciences, The University of Adelaide(计算机与数学科学学院,阿德莱德大学)