Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement

Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement

Contents

1. Abstract

Abstract

$D^3VAE$

• Address the TS forecasting problem with generative modeling
• Bidirectional VAE (BVAE) equipped with diffusion, denoise, and disentanglement
• Coupled diffusion probabilistic model
  • To augment the TS data without increasing the aleatoric uncertainty & implement a more tractable inference process with BVAE
• Propose to adapt and integrate the multiscale denoising score matching into the diffusion process for TS forecasting
• To enhance the interpretability and stability of the prediction, treat the latent variable in a multivariate manner and disentangle them on top of minimizing total correlation