9 August 2017 in Journal of Advanced Transportation
Accurate traffic flow prediction is increasingly essential for successful traffic modeling, operation, and management. Traditional data driven traffic flow prediction approaches have largely assumed restrictive (shallow) model architectures and do not leverage the large amount of environmental data available. Inspired by deep learning methods with more complex model architectures and effective data mining capabilities, this paper introduces the deep belief network (DBN) and long short-term memory (LSTM) to predict urban traffic flow considering the impact of rainfall. The rainfall-integrated DBN and LSTM can learn the features of traffic flow under various rainfall scenarios. Experimental results indicate that, with the consideration of additional rainfall factor, the deep learning predictors have better accuracy than existing predictors and also yield improvements over the original deep learning models without rainfall input. Furthermore, the LSTM can outperform the DBN to capture the time series characteristics of traffic flow data.
11 May 2018 in IEEE Transactions on Intelligent Transportation Systems
Road networks are extremely vulnerable to cascading failure caused by traffic accidents or anomalous events. Therefore, accurate identification of critical nodes, whose failure may cause a dramatic reduction in the road network transmission efficiency, is of great significance to traffic management and control schemes. However, none of the existing approaches can locate city-wide critical nodes in real road networks. In this paper, we propose a novel data-driven framework to rank node importance through mining from comprehensive vehicle trajectory data, instead of analysis solely on the topology of the road network. In this framework, we introduce a trip network modeled by a tripartite graph to characterize the dynamics of the road network. Furthermore, we present two algorithms, integrating the origin-destination entropy with flow (ODEF) algorithm and the crossroad-rank (CRRank) algorithm, to better exploit the information included in the tripartite graph and to effectively assess the node importance. ODEF absorbs the idea of the information entropy to evaluate the centrality of a node and to calculate its importance rating by integrating its centrality with the traffic flow. CRRank is a ranking algorithm based on eigenvector centrality that captures the mutual reinforcing relationships among the OD-pair, path, and intersection. In addition to the factors considered in ODEF, CRRank considers the irreplaceability of a node and the spatial relationships between neighboring nodes. We conduct a synthetic experiment and a real case study based on a real-world dataset of taxi trajectories. Experiments verify the utility of the proposed algorithms.
25 September 2019 in IEEE Transactions on Intelligent Transportation Systems
Anomaly detection in road networks is vital for traffic management and emergency response. However, existing approaches do not directly address multiple anomaly types. We propose a tensor-based spatio-temporal model for detecting multiple types of anomalies in road networks. First, we represent network traffic data as a 3rd-order tensor. Next, we acquire spatial and multi-scale temporal patterns of traffic variations via a novel, computationally efficient tensor factorization algorithm: sliding window tensor factorization. Then, from the factorization results, we can identify different anomaly types by measuring deviations from different spatial and temporal patterns. Finally, we discover path-level anomalies by formulating anomalous path inference as a linear program that solves for the best matched paths of anomalous links. We evaluate the proposed methods via both synthetic experiments and case studies based on a real-world vehicle trajectory dataset, demonstrating advantages of our approach over baselines.
20 March 2018 in IEEE Transactions on Computational Social Systems
In social networks, link establishment among the users is affected by complex factors. In this paper, we try to investigate the internal and external factors that affect the formation of links and propose a three-level hidden Bayesian link prediction model by integrating the user behavior as well as user relationships to link prediction. First, based on the user multiple interest characteristics, a latent Dirichlet allocation (LDA) traditional text modeling method is applied into user behavior modeling. Taking the advantage of LDA topic model in dealing with the problem of polysemy and synonym, we can mine user latent interest distribution and analyze the effects of internal driving factors. Second, owing to the power-law characteristics of user behavior, LDA is improved by Gaussian weighting. In this way, the negative impact of the interest distribution to the high-frequency users can be reduced and the expression ability of interests can be enhanced. Furthermore, taking the impact of common neighbor dependencies in link establishment, the model can be extended with hidden naive Bayesian algorithm. By quantifying the dependencies between common neighbors, we can analyze the effects of external driving factors and combine internal driving factors to link prediction. Experimental results indicate that the model can not only mine user latent interest distribution but also can improve the performance of link prediction effectively.
1 January 2015 in Mathematical Problems in Engineering
In order to improve offline map matching accuracy of uncertain GPS trajectories, a map matching algorithm based on conditional random fields (CRF) and route preference mining is proposed. In this algorithm, road offset distance and the temporal-spatial relationship between the sampling points are used as features of GPS trajectory in a CRF model, which integrates the temporal-spatial context information flexibly. The driver route preference is also used to bolster the temporal-spatial context when a low GPS sampling rate impairs the resolving power of temporal-spatial context in CRF, allowing the map matching accuracy of uncertain GPS trajectories to get improved significantly. The experimental results show that our proposed algorithm is more accurate than existing methods, especially in the case of a low-sampling-rate.
1 August 2017 in Information Sciences