Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement

时间序列预测是一项具有重要意义的研究课题。然而，现实世界的时间序列数据通常记录在较短的时间段内，这导致深度模型与有限且有噪声的时间序列之间存在较大差距。本文提出用生成式建模来解决时间序列预测问题，并提出一种具有扩散、去噪和解缠的双向变分自编码器(BVAE)，即D3VAE。具体地，提出一种耦合扩散概率模型，在不增加任意不确定性的情况下对时间序列数据进行增强，并利用BVAE实现更易于处理的推理过程。为了确保生成的序列朝着真实目标移动，进一步提出将多尺度去噪得分匹配自适应并集成到时间序列预测的扩散过程中。此外，为了增强预测的可解释性和稳定性，以多元的方式处理潜变量，并在最小化总相关性的基础上解缠它们。在合成数据和真实数据上的广泛实验表明，D3VAE以显著的优势优于竞争算法。

•我们提出了一种耦合扩散概率模型，旨在降低时间序列的任意不确定性，提高生成模型的泛化能力。
•我们将多尺度去噪评分匹配整合到耦合扩散过程中，以提高生成结果的准确性。
•我们对生成模型的潜在变量进行解纠缠，以提高时间序列预测的可解释性。
•在合成数据集和真实数据集上的大量实验表明，D3VAE的性能优于竞争基线，

Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement相关推荐

愉快的学习就从翻译开始吧_0-Time Series Forecasting with the Long Short-Term Memory Network in Python
The Long Short-Term Memory recurrent neural network has the promise of learning long sequences of ob ...
论文阅读：Recurrent Neural Networks for Time Series Forecasting Current Status and Future Directions
typora-copy-images-to: ./ Recurrent Neural Networks for Time Series Forecasting: Current Status and ...
Towards Spatio-Temporal Aware Traffic Time Series Forecasting
Towards Spatio-Temporal Aware Traffic Time Series Forecasting 摘要:由于复杂的时空动态,不同地点的时间序列往往具有不同的模式,交通时间序列 ...
论文笔记：Do We Really Need Deep Learning Models for Time Series Forecasting?
Do We Really Need Deep Learning Models for Time Series Forecasting? Elsayed S, Thyssens D, Rashed A, ...
Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks
Connecting the Dots: Multivariate Time Series Forecasting with Graph Neural Networks 摘要:长期以来,多元时间序列建 ...
时间序列预测分析（2）How to Develop a Skillful Machine Learning Time Series Forecasting Model
文章目录 Process Overview How to Use This Process 1. Define Problem 分析问题 2. Design Test Harness 设计评估工具常 ...
【翻译1】Multivariate Time Series Forecasting with LSTMs in Keras——PM2.5
LSTM-PM2.5预测例程 1. Air Pollution Forecasting 2. Basic Data Preparation 3. Multivariate LSTM Forecast ...
【论文精读】Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting
[论文精读]Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting 针对未来的一个多步 ...
N-HiTS: Neural Hierarchical Interpolation for Time Series Forecasting
N-HiTS: Neural Hierarchical Interpolation for Time Series Forecasting 神经预测的最新进展加速了大规模预测系统性能的提高.然而,长期 ...

Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement

Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement相关推荐

最新文章

热门文章