【1. 概要】 论文针对的是时序预测问题(Time series forecasting,TSF),根据时间序列的特点创新性地提出了一个多层的神经网络框架sample convolution and interaction network(SCINet)用于时序预测。模型在多个数据集上都展示了其准确率上的优越性,且时间成本相对其他模型(如时序卷积网络TCN)也更低。本篇论文工作包含以下...
文章提出了一种新的时间序列预测模型TimeCNN,通过引入时间点独立的卷积核,精炼跨变量交互,以更好地捕捉多变量时间序列中复杂的动态关系,从而在多个实际数据集上实现了优于现有模型的预测性能和计算效率。 论文题目:TimeCNN: Refining Cross-Variable Interaction on Time Point for Time Series Forecasting 论文链接:http...
2.1 Models for Time Series Forecasting 由于时间序列预测的巨大重要性,各种模型已经得到很好的发展。许多时间序列预测方法都是从经典工具开始的[38,10]。ARIMA[7,6]通过差分将非平稳过程转化为平稳过程来解决预测问题。在序列预测中也引入了滤波方法[24,12]。此外,利用循环神经网络(RNNs)模型对时间序列的时间相关性...
本文基于前期介绍的风速数据(文末附数据集),介绍一种多特征变量序列预测模型CNN-LSTM,以提高时间序列数据的预测性能。该数据集一共有天气、温度、湿度、气压、风速等九个变量,通过滑动窗口制作数据集,利用多变量来预测风速。 LSTF(Long Sequence Time-Series Forecasting)问题是指在时间序列预测中需要处理长序列的情况...
2023年J. P. Morgan AI Research发布《Financial Time Series Forecasting using CNN and Transformer》，...
,series(2:end)',layers,options); % 使用 LSTM 模型进行预测 futureValues = predict(net, series...
英文标题:Financial Time Series Forecasting using CNN and Transformer中文摘要:本文提出了通过使用卷积神经网络和 Transformers 来捕捉时间序列中的短期和长期依赖,并用于预测股票价格变化,与传统的统计和深度学习方法相比,实验结果表明该方法取得了成功。英文摘要:Time series forecasting is important across various domains...
However, Transformers have their limitations when training on small datasets because of their lack in necessary inductive bias for time series forecasting, and do not show significant benefits in short-time step forecasting as well as that in long-time step as the continuity of sequence is not ...
forspeech recognitionand natural language processing tasks, such as text summarization, machine translation and speech analysis. Example use cases for RNNs include generating textual captions for images,forecasting time series datasuch as sales or stock prices, andanalyzing user sentimentin social media ...
the Mean AbsoluteScaled Error (MASE)and Root Mean Square Error (RMSE)have been increased by 89.64% and 61.73% respectively.Key words:attention mechanism; Convolution Neural Network(CNN); Long Short-Term Memory Network(LSTM);time series; load forecasting基金项目:中国科学院战略性先导科技专项(No.XDA...