Discrete Graph Structure Learning for Forecasting Multiple Time Series

By Chao Shang, Jie Chen, Jinbo Bi. International Conference on Learning Representations, 2021

377 citations recorded by the source corpus.

Time series forecasting is an extensively studied subject in statistics, economics, and computer science. Exploration of the correlation and causation among the variables in a multivariate time series shows promise in enhancing the performance of a time series model. When using deep neural networks as forecasting models, we hypothesize that exploiting the pairwise information among multiple (multivariate) time series also improves their forecast. If an explicit graph structure is known, graph neural networks (GNNs) have been demonstrated as powerful tools to exploit the structure. In this work, we propose learning the structure simultaneously with the GNN if the graph is unknown. We cast the problem as learning a probabilistic graph model through optimizing the mean performance over the graph distribution. The distribution is parameterized by a neural network so that discrete graphs can be sampled differentiably through reparameterization. Empirical evaluations show that our method is simpler, more efficient, and better performing than a recently proposed bilevel learning approach for graph structure learning, as well as a broad array of forecasting models, either deep or non-deep learning based, and graph or non-graph based.

Subfields: Artificial Intelligence, Building and Construction, Signal Processing

View Paper: https://www.semanticscholar.org/paper/401930b2dc738a5a67d136bc9a2d04461c5bf93a

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Building and Construction

In Signal Processing

Citations in the corpus

Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting (2022, background): “Many researchers have made persistent efforts to design powerful graph convolutions [2, 20, 22], or to reduce reliance on the pre-defined graph structure [11, 13, 17].”
Pre-training Enhanced Spatial-temporal Graph Neural Network for Multivariate Time Series Forecasting (2022, methodology): “In fact, most models use historical data in a small window to make predictions, e.g., use the past twelve time steps (one hour) to predict the future twelve time steps [20, 29, 35, 36, 41].”
A Survey on Graph Neural Networks for Time Series: Forecasting, Classification, Imputation, and Anomaly Detection (2023, background): “More recent studies, such as GTS [52] and RGSL [72], share similar designs but primarily emphasize different graph structure learning mechanisms.”
Spatio-Temporal Graph Neural Networks for Predictive Learning in Urban Computing: A Survey (2023, methodology): “Feature initialization-based approaches aim to construct adaptive graph structure learning based on the given inputs or the hidden states [58], [66], [94], [162], [163], [164].”
Spatio-Temporal Adaptive Embedding Makes Vanilla Transformer SOTA for Traffic Forecasting (2023, methodology): “Researchers have devoted considerable effort to developing fancy and complicated models for traffic forecasting, such as novel graph convolutions [2, 3, 9–11, 13, 18, 21, 26, 29, 31, 37, 41], learning graph structures [15, 23, 32, 34, 35, 40], efficient attention mechanisms [5, 20, 33, 43, 44], and ”
Spatio-Temporal Meta-Graph Learning for Traffic Forecasting (2022, neutral): “However, due to the reproducibility problem ( ⋆ in Table 2), we report the results of GMAN, GTS, and PM-MemNet either from our own experiments or other literature (e.g., GMAN on METR-LA (Shao et al. 2022)).”
Deep Learning for Time Series Forecasting: Tutorial and Literature Survey (2020, methodology): “As of now, two main strategies have been proposed to learn the node embeddings, either simply by gradient descent [43, 193] or by taking representation from the time series [102, 163], with the latter approach to seemingly yielding better results.”