Spatial-Temporal Hypergraph Self-Supervised Learning for Crime Prediction

Li, Zhonghang; Huang, Chao; Xia, Lianghao; Xu, Yong; Pei, Jian

doi:10.1109/icde53745.2022.00269

Cited by 34 publications

(25 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Contisciani et al (2022) proposed a statistical inference methods named Hypergraph-MT to identify communities with higher-order interactions and infer missing hyperedges. A spatiotemporal self-supervised hypergraph learning was proposed for city crime prediction (Li et al, 2022), which enables crime data enhancement and city-wide crime characterization. Since hypergraph learning has produced excellent outcomes for modeling data higher-order relationships, scholars have constructed variant models based on hypergraph structure for traffic prediction (Wang and Zhu, 2022).…”

Section: Hypergraph Learningmentioning

confidence: 99%

“…From a temporal perspective, we have similar conclusions from Section 3.2 that the occurrence of traffic accidents is cyclical and dependent on longer time scales. Based on the aforementioned issues, we employ a hypergraph learning architecture aspired by (Li et al, 2022), which consists of a long temporal encoder and a regional hypergraph spatial model for capturing cross-regional global dependencies at global scales.…”

Section: Global Dynamic Hypergraph Networkmentioning

confidence: 99%

See 1 more Smart Citation

Sparse Spatio-Temporal Dynamic Hypergraph Learning for Traffic Accident Prediction

Cui¹,

Yang

Abdel-Aty

2023

Preprint

View full text Add to dashboard Cite

Traffic accidents have become one of the biggest public health safety matters, which has raised many concerns from citizens and city managers. Accurate traffic accident prediction can not only assist the government in making decisions in advance but also enhance public trust in public safety. Conventional spatio-temporal prediction models, limited by the skewed distributions and sparse labels of traffic accident occurrence, are prone to overfitting. Inspired by hypergraph learning and self-supervised learning, this paper suggests a sparse spatio-temporal dynamic hypergraph learning (SST-DHL) framework to capture the higher-order dependencies in sparse traffic accidents. Specifically, a multi-view spatio-temporal convolution block is employed first to capture the local spatio-temporal correlation and inherent semantics of traffic accidents. Then we propose a cross-regional dynamic hypergraph learning model to capture global spatio-temporal dependencies beneath the entire urban landscape. In addition, a two-supervised self-learning paradigm is intended to strengthen the representation of sparse traffic occurrences by regional self-identification, which can capture local and global spatio-temporal traffic patterns. The proposed model is applicable to most sparse datasets for traffic forecasts. Extensive experiments was conducted on two heterogeneous accident datasets from New York City and London, and the results shows an average improvements of 7.21%-23.09% at different sparsity levels compared to the optimal baselines. More importantly, the proposed SST-DHL improves the interpretability of model results, which demonstrates that hypergraph learning can efficiently capture the complex higher-order spatio-temporal dependencies among different traffic accident instances.

show abstract

Section: Hypergraph Learningmentioning

confidence: 99%

Section: Global Dynamic Hypergraph Networkmentioning

confidence: 99%

Sparse Spatio-Temporal Dynamic Hypergraph Learning for Traffic Accident Prediction

Cui¹,

Yang

Abdel-Aty

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Such diverse spatio-temporal data drives the need for effective spatio-temporal prediction frameworks for various urban sensing applications, such as traffic analysis [28], human mobility behavior modeling [15], and citywide crime prediction [8]. For instance, motivated by the opportunities of building machine learning and big data driven intelligent cities, the discovered human trajectory patterns can help to formulate better urban planning mechanisms [3], or understanding the dynamics of crime occurrences is useful for reducing crime rate [12,26].…”

Section: Introductionmentioning

confidence: 99%

Automated Spatio-Temporal Graph Contrastive Learning

Zhang

Huang

Xia

et al. 2023

Proceedings of the ACM Web Conference 2023

View full text Add to dashboard Cite

Among various region embedding methods, graph-based region relation learning models stand out, owing to their strong structure representation ability for encoding spatial correlations with graph neural networks. Despite their effectiveness, several key challenges have not been well addressed in existing methods: i) Data noise and missing are ubiquitous in many spatio-temporal scenarios due to a variety of factors. ii) Input spatio-temporal data (e.g., mobility traces) usually exhibits distribution heterogeneity across space and time. In such cases, current methods are vulnerable to the quality of the generated region graphs, which may lead to suboptimal performance. In this paper, we tackle the above challenges by exploring the Automated Spatio-Temporal graph contrastive learning paradigm (AutoST) over the heterogeneous region graph generated from multi-view data sources. Our AutoST framework is built upon a heterogeneous graph neural architecture to capture the multi-view region dependencies with respect to POI semantics, mobility flow patterns and geographical positions. To improve the robustness of our GNN encoder against data noise and distribution issues, we design an automated spatio-temporal augmentation scheme with a parameterized contrastive view generator. AutoST can adapt to the spatio-temporal heterogeneous graph with multiview semantics well preserved. Extensive experiments for three downstream spatio-temporal mining tasks on several real-world datasets demonstrate the significant performance gain achieved by our AutoST over a variety of baselines. The code is publicly available at https://github.com/HKUDS/AutoST.

show abstract

“…The self-attention mechanism has also been employed and shown to be effective in modeling spatio-temporal dependency [9,57,58]. On the other hand, in the context of crime prediction, recurrent attentive networks are utilized to model complicated spatio-temporal crime patterns [15], while Hypergraph Neural Networks [47] and Self-Supervised Learning [26] have been employed to learn global spatio-temporal dependencies and address specific challenges in learning crime patterns. .…”

Section: Introductionmentioning

confidence: 99%

“…In addition to the limited number of samples, data missing problems often occur in real-world spatiotemporal applications due to various reasons, such as sensor failure in traffic scenarios and data privacy in epidemic forecasting. ii) Data Sparsity is another issue in some spatio-temporal forecasting tasks, such as crime prediction [26] and epidemic forecasting [42]. In these cases, the data of each fine-grained region or sensor can be sparse along the temporal dimension when compared to the whole urban space.…”

Section: Introductionmentioning

confidence: 99%

Spatio-Temporal Meta Contrastive Learning

Tang,

Xia,

et al. 2023

Proceedings of the 32nd ACM International Conference on Information and Knowledge Management

View full text Add to dashboard Cite

Spatio-temporal prediction is crucial in numerous real-world applications, including traffic forecasting and crime prediction, which aim to improve public transportation and safety management. Many state-of-the-art models demonstrate the strong capability of spatiotemporal graph neural networks (STGNN) to capture complex spatio-temporal correlations. However, despite their effectiveness, existing approaches do not adequately address several key challenges. Data quality issues, such as data scarcity and sparsity, lead to data noise and a lack of supervised signals, which significantly limit the performance of STGNN. Although recent STGNN models with contrastive learning aim to address these challenges, most of them use pre-defined augmentation strategies that heavily depend on manual design and cannot be customized for different Spatio-Temporal Graph (STG) scenarios. To tackle these challenges, we propose a new spatio-temporal contrastive learning (CL4ST) framework to encode robust and generalizable STG representations via the STG augmentation paradigm. Specifically, we design the meta view generator to automatically construct node and edge augmentation views for each disentangled spatial and temporal graph in a data-driven manner. The meta view generator employs meta networks with parameterized generative model to customize the augmentations for each input. This personalizes the augmentation strategies for every STG and endows the learning framework with spatio-temporal-aware information. Additionally, we integrate a unified spatio-temporal graph attention network with the proposed meta view generator and two-branch graph contrastive learning paradigms. Extensive experiments demonstrate that our CL4ST significantly improves performance over various state-of-the-art baselines in traffic and crime prediction. Our model implementation is available at the link: https://github.com/HKUDS/CL4ST.

show abstract

Spatial-Temporal Hypergraph Self-Supervised Learning for Crime Prediction

Cited by 34 publications

References 46 publications

Sparse Spatio-Temporal Dynamic Hypergraph Learning for Traffic Accident Prediction

Sparse Spatio-Temporal Dynamic Hypergraph Learning for Traffic Accident Prediction

Automated Spatio-Temporal Graph Contrastive Learning

Spatio-Temporal Meta Contrastive Learning

Contact Info

Product

Resources

About