Abstract:The non-stationary nature of real-world Multivariate Time Series (MTS) data presents forecasting models with a formidable challenge of the time-variant distribution of time series, referred to as distribution shift. Existing studies on the distribution shift mostly adhere to adaptive normalization techniques for alleviating temporal mean and covariance shifts or time-variant modeling for capturing temporal shifts. Despite improving model generalization, these normalization-based methods often assume a time-invariant transition between outputs and inputs but disregard specific intra-/inter-series correlations, while time-variant models overlook the intrinsic causes of the distribution shift. This limits model expressiveness and interpretability of tackling the distribution shift for MTS forecasting. To mitigate such a dilemma, we present a unified Probabilistic Graphical Model to Jointly capturing intra-/inter-series correlations and modeling the time-variant transitional distribution, and instantiate a neural framework called JointPGM for non-stationary MTS forecasting. Specifically, JointPGM first employs multiple Fourier basis functions to learn dynamic time factors and designs two distinct learners: intra-series and inter-series learners. The intra-series learner effectively captures temporal dynamics by utilizing temporal gates, while the inter-series learner explicitly models spatial dynamics through multi-hop propagation, incorporating Gumbel-softmax sampling. These two types of series dynamics are subsequently fused into a latent variable, which is inversely employed to infer time factors, generate final prediction, and perform reconstruction. We validate the effectiveness and efficiency of JointPGM through extensive experiments on six highly non-stationary MTS datasets, achieving state-of-the-art forecasting performance of MTS forecasting.

Recurrent Interpolants for Probabilistic Time Series Prediction

Latent Diffusion Transformer for Probabilistic Time Series Forecasting

Enhancing Time Series Forecasting: A Hierarchical Transformer with Probabilistic Decomposition Representation

Interacting Diffusion Processes for Event Sequence Forecasting

Diffusion-TS: Interpretable Diffusion for General Time Series Generation

History Marginalization Improves Forecasting in Variational Recurrent Neural Networks

Multivariate Probabilistic Time Series Forecasting via Conditioned Normalizing Flows

Non-autoregressive Conditional Diffusion Models for Time Series Prediction

Probabilistic Time Series Modeling with Decomposable Denoising Diffusion Model

Predict, Refine, Synthesize: Self-Guiding Diffusion Models for Probabilistic Time Series Forecasting

Probabilistic Decomposition Transformer for Time Series Forecasting

TS-Diffusion: Generating Highly Complex Time Series with Diffusion Models

Probabilistic Forecasting with Stochastic Interpolants and Föllmer Processes

Robust Multivariate Time Series Forecasting against Intra- and Inter-Series Transitional Shift

Probabilistic Forecasting of Irregular Time Series via Conditional Flows

Non-parametric Probabilistic Time Series Forecasting via Innovations Representation

Stochastic Recurrent Neural Network for Multistep Time Series Forecasting

Stochastic Diffusion: A Diffusion Probabilistic Model for Stochastic Time Series Forecasting

REDI: Recurrent Diffusion Model for Probabilistic Time Series Forecasting

Deep probabilistic time series forecasting using augmented recurrent input for dynamic systems

Probabilistic Learning of Multivariate Time Series with Temporal Irregularity