Adaptive Graph Convolutional Recurrent Network for Traffic Forecasting
By Lei Bai, Lina Yao, Can Li, Xianzhi Wang, Can Wang. Neural Information Processing Systems, 2020
2,016 citations recorded by the source corpus.
Modeling complex spatial and temporal correlations in the correlated time series data is indispensable for understanding the traffic dynamics and predicting the future status of an evolving traffic system. Recent works focus on designing complicated graph neural network architectures to capture shared patterns with the help of pre-defined graphs. In this paper, we argue that learning node-specific patterns is essential for traffic forecasting while the pre-defined graph is avoidable. To this end, we propose two adaptive modules for enhancing Graph Convolutional Network (GCN) with new capabilities: 1) a Node Adaptive Parameter Learning (NAPL) module to capture node-specific patterns; 2) a Data Adaptive Graph Generation (DAGG) module to infer the inter-dependencies among different traffic series automatically. We further propose an Adaptive Graph Convolutional Recurrent Network (AGCRN) to capture fine-grained spatial and temporal correlations in traffic series automatically based on the two modules and recurrent networks. Our experiments on two real-world traffic datasets show AGCRN outperforms state-of-the-art by a significant margin without pre-defined graphs about spatial connections.
Subfields: Artificial Intelligence, Building and Construction, Statistical and Nonlinear Physics
View Paper: https://www.semanticscholar.org/paper/8b163b75a6b833911c4e958f8bd52124205382ec
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 Statistical and Nonlinear Physics
Citations in the corpus
- Spatial-Temporal Identity: A Simple yet Effective Baseline for Multivariate Time Series Forecasting (2022, methodology): “HI MAE 42.33 42.35 42.37 42.36 49.02 49.03 49.06 49.04 34.55 34.57 34.59 34.57 3.06 3.06 3.05 3.06 92.44 92.58 92.79 92.58 RMSE 61.64 61.66 61.67 61.66 71.16 71.18 71.20 71.18 50.41 50.43 50.44 50.43 7.05 7.05 7.03 7.04 167.00 167.05 167.21 167.07 MAPE 29.90% 29.92% 29.96% 29.92% 22.73% 22.75% 22.79”
- Dynamic Graph Convolutional Recurrent Network for Traffic Prediction: Benchmark and Solution (2021, methodology): “Graph WaveNet, MTGNN and AGCRN benefit a lot from the self-adaptive adjacency matrix whose parameters are learnable and can be updated with back propagation algorithm.”
- Frequency-domain MLPs are More Effective Learners in Time Series Forecasting (2023, methodology): “…Neural Network-based methods (e.g., DeepAR [9], LSTNet [10]), Convolution Neural Network-based meth-ods (e.g., TCN [11], SCINet [12]), Transformer-based methods (e.g., Informer [13], Autoformer [14]), and Graph Neural Network-based methods (e.g., MTGNN [15], StemGNN [16], AGCRN [17]), etc.”
- Graph Neural Network for Traffic Forecasting: A Survey (2021, background): “Problem Relevant Studies Road Traffic Flow [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62],”
- Pre-training Enhanced Spatial-temporal Graph Neural Network for Multivariate Time Series Forecasting (2022, background): “DCRNN [20], ST-MetaNet [26], AGCRN [2], and TGCN [40] combine diffusion convolutional networks and recurrent neural networks [6, 32] with their variants.”
- FourierGNN: Rethinking Multivariate Time Series Forecasting from a Pure Graph Perspective (2023, neutral): “For example, AGCRN [2] use a GCN [14] and a GRU [21], GraphWaveNet [17] use a GCN and a TCN [8], etc.”
- Dynamic and Multi-faceted Spatio-temporal Deep Learning for Traffic Speed Forecasting (2021, methodology): “Most of the recent research has focused on geometric spatio-temporal learning [1, 14, 19, 26, 27, 31, 32, 35].”
- Spatio-Temporal Adaptive Embedding Makes Vanilla Transformer SOTA for Traffic Forecasting (2023, methodology): “We also consider STGNNs such as GWNet [34], DCRNN [18], AGCRN [1], STGCN [37], GTS [23], and MTGNN [33], which employ the embeddings shown in Figure 1(a).”
- PDFormer: Propagation Delay-aware Dynamic Long-range Transformer for Traffic Flow Prediction (2023, background): “Later, thanks to the powerful ability to model graph data, graph neural networks (GNNs) were widely used for traffic prediction (Chen et al. 2020; Bai et al. 2020; Zhang et al. 2021; Oreshkin et al. 2021; Ye et al. 2021; Han et al. 2021; Ji et al. 2022; Liu et al. 2022).”
- Spatio-Temporal Meta-Graph Learning for Traffic Forecasting (2022, neutral): “Here we adopt the canonical formulation (Wu et al. 2019; Bai et al. 2020; Wang et al. 2022) for spatio-temporal graph learning, namely adaptive graph (in Figure 1), denoted by: where ˜ P is derived by random walk normalizing the non-negative part of matrix product of trainable node embedding E ∈ R…”
- SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction (2021, background): “This is because, on the one hand, Autoformer focuses on modeling seasonal patterns and conducts self-attention at the sub-series level (instead of the raw data), which is much better in extracting long-term temporal patterns than vanilla Transformer-based methods.”
- Spatio-Temporal Self-Supervised Learning for Traffic Flow Prediction (2022, background): “To further explore why the embeddings obtained by ST-SSL can deliver more accurate traffic prediction than AGCRN, we visualize them on BJTaxi by t-SNE (Van der Maaten and Hinton 2008).”
- Graph Neural Controlled Differential Equations for Traffic Forecasting (2021, methodology): “However, it is outperformed by AGCRN and Z-GCNETs for PeMSD7.”