Inferring Intersection Traffic Patterns With Sparse Video Surveillance Information: An ST-GAN Method

Wang, Pengkun; Zhu, Chaochao; Wang, Xu; Zhou, Zhengyang; Wang, Guang; Wang, Yang

doi:10.1109/tvt.2022.3176653

Cited by 11 publications

(3 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the conspicuous progress of data mining techniques, spatio-temporal prediction has unprecedentedly facilitated today's society, such as traffic state modeling (Zhang, Zheng, and Qi 2017;Wang et al 2022;Zhou et al 2020;Xu et al 2016), urban crime prediction (Zhao et al 2022b;Zhao and Tang 2017a,b), next point-of-interest recommendation (Guo et al 2016;Cui et al 2021), etc.Spatio-temporal prediction aims to model spatial and temporal patterns from historical spatio-temporal data, and predict the future states.…”

Section: Introductionmentioning

confidence: 99%

AutoSTL: Automated Spatio-Temporal Multi-Task Learning

Zhang

Zhao

Hao

et al. 2023

AAAI

View full text Add to dashboard Cite

Spatio-temporal prediction plays a critical role in smart city construction. Jointly modeling multiple spatio-temporal tasks can further promote an intelligent city life by integrating their inseparable relationship. However, existing studies fail to address this joint learning problem well, which generally solve tasks individually or a fixed task combination. The challenges lie in the tangled relation between different properties, the demand for supporting flexible combinations of tasks and the complex spatio-temporal dependency. To cope with the problems above, we propose an Automated Spatio-Temporal multi-task Learning (AutoSTL) method to handle multiple spatio-temporal tasks jointly. Firstly, we propose a scalable architecture consisting of advanced spatio-temporal operations to exploit the complicated dependency. Shared modules and feature fusion mechanism are incorporated to further capture the intrinsic relationship between tasks. Furthermore, our model automatically allocates the operations and fusion weight. Extensive experiments on benchmark datasets verified that our model achieves state-of-the-art performance. As we can know, AutoSTL is the first automated spatio-temporal multi-task learning method.

show abstract

Section: Introductionmentioning

confidence: 99%

AutoSTL: Automated Spatio-Temporal Multi-Task Learning

Zhang

Zhao

Hao

et al. 2023

AAAI

View full text Add to dashboard Cite

show abstract

“…Trajectory data is an important basis for the rapid development of intelligent transportation applications. For example, applications such as travel estimation [1], urban planning [2], and traffic spatiotemporal prediction [3]- [6] all require a large amount of trajectory data for support. The performance of these applications depends on the trajectory sampling rate, because the low sampling rate will lead to an increase in the distance between two adjacent trajectory points, thereby increasing the uncertainty of the trajectory [7].…”

Section: Introductionmentioning

confidence: 99%

Road-constrained trajectory recovery via BERT-based multi-task learning

Shi

Yang

Wang

2023

International Conference on Intelligent Systems, Communications, and Computer Networks (ISCCN 2023)

View full text Add to dashboard Cite

With the increasing popularity of GPS modules, the development of intelligent transportation applications based on trajectory data is also accelerating, such as traffic forecasting, vehicle navigation, and travel time estimation. These urban applications all require a large number of highly sampled trajectories as a basis. However, due to energy constraints and environmental interference, a large number of trajectories are collected at low sampling rates. In order to solve the uncertainty problem brought by low sampling rate trajectories, many trajectory recovery algorithms have been proposed. Currently, many existing works are based on sequence-to-sequence networks for trajectory recovery, that is, encoders are used to encode trajectories, and decoders are used to recover missing trajectory points. However, these works do not consider road network information and only use grid numbers to recover trajectories. Therefore, we propose a BERTbased multi-task learning framework, namely BTTrajRec, for trajectory recovery. BTTrajRec incorporates road information when encoding trajectories, and captures the bidirectional relationship between trajectory points, which can better characterize trajectory data. In addition, a constraint mask matrix is used to improve the recovery accuracy. We used multiple taxi data sets in the real world to verify the validity of the proposed method.

show abstract

“…Therefore, the nearest statuses play a significant role in forecasting as they can provide key informative knowledge for status estimations on following consecutive steps. Even for those sparse spatiotemporal learning efforts where sensors are sparsely deployed in spatial domain, researchers can still take the status of spatially neighboring sensors as proxy [10] or generate real-time data with well-designed discriminators [11]. Unfortunately, based on above analysis, these solutions are actually an interpolation strategy that takes great advantage of nearest observations in both temporal and spatial perspectives, and are incapable of dealing with the serial observation missing issue in our task.…”

Section: Introductionmentioning

confidence: 99%

Towards Learning in Grey Spatiotemporal Systems: A Prophet to Non-consecutive Spatiotemporal Dynamics

Zhou¹,

Kuo²,

Sun³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Spatiotemporal forecasting is an imperative topic in data science due to its diverse and critical applications in smart cities. Existing works mostly perform consecutive predictions of following steps with observations completely and continuously obtained, where nearest observations can be exploited as key knowledge for instantaneous status estimation. However, the practical issues of early activity planning and sensor failures elicit a brand-new task, i.e., non-consecutive forecasting. In this paper, we define spatiotemporal learning systems with missing observation as Grey Spatiotemporal Systems (G2S) and propose a Factor-Decoupled learning framework for G2S (FDG2S), where the core idea is to hierarchically decouple multi-level factors and enable both flexible aggregations and disentangled uncertainty estimations. Firstly, to compensate for missing observations, a generic semantic-neighboring sequence sampling is devised, which selects representative sequences to capture both periodical regularity and instantaneous variations. Secondly, we turn the predictions of non-consecutive statuses into inferring statuses under expected combined exogenous factors. In particular, a factor-decoupled aggregation scheme is proposed to decouple factor-induced predictive intensity and region-wise proximity by two energy functions of conditional random field. To infer region-wise proximity under flexible factor-wise combinations and enable dynamic neighborhood aggregations, we further disentangle compounded influences of exogenous factors on regionwise proximity and learn to aggregate them. Given the inherent incompleteness and critical applications of G2S, a DisEntangled Uncertainty Quantification is put forward, to identify two types of uncertainty for reliability guarantees and model interpretations.Extensive experiments demonstrate that our solution can promote the performance by at least 8.50% on early planning and 2.01%-18.00% on the sensor failure setting.

show abstract

Inferring Intersection Traffic Patterns With Sparse Video Surveillance Information: An ST-GAN Method

Cited by 11 publications

References 41 publications

AutoSTL: Automated Spatio-Temporal Multi-Task Learning

AutoSTL: Automated Spatio-Temporal Multi-Task Learning

Road-constrained trajectory recovery via BERT-based multi-task learning

Towards Learning in Grey Spatiotemporal Systems: A Prophet to Non-consecutive Spatiotemporal Dynamics

Contact Info

Product

Resources

About