Learning to Route with Sparse Trajectory Sets

Guo, Chenjuan; Yang, Bin; Hu, Jilin; Jensen, Christian S.

doi:10.1109/icde.2018.00100

Cited by 51 publications

(26 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Experiments on two real world data sets show promising results. In the future, it is of interest to verify the proposed models on time series from different domains such as transportation [3,4,7] and to study the scalability of the proposed models for large time series data, e.g., by exploring parallel computing frameworks [24,25].…”

Section: Discussionmentioning

confidence: 99%

Correlated Time Series Forecasting using Multi-Task Deep Neural Networks

Cirstea

Micu

Muresan

et al. 2018

Proceedings of the 27th ACM International Conference on Information and Knowledge Management

Self Cite

View full text Add to dashboard Cite

Cyber-physical systems often consist of entities that interact with each other over time. Meanwhile, as part of the continued digitization of industrial processes, various sensor technologies are deployed that enable us to record time-varying attributes (a.k.a., time series) of such entities, thus producing correlated time series. To enable accurate forecasting on such correlated time series, this paper proposes two models that combine convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The first model employs a CNN on each individual time series, combines the convoluted features, and then applies an RNN on top of the convoluted features in the end to enable forecasting. The second model adds additional auto-encoders into the individual CNNs, making the second model a multi-task learning model, which provides accurate and robust forecasting. Experiments on two real-world correlated time series data set suggest that the proposed two models are effective and outperform baselines in most settings. This report extends the paper "Correlated Time Series Forecasting using Multi-Task Deep Neural Networks, " to appear in ACM CIKM 2018, by providing additional experimental results.

show abstract

Section: Discussionmentioning

confidence: 99%

Correlated Time Series Forecasting using Multi-Task Deep Neural Networks

Cirstea

Micu

Muresan

et al. 2018

Proceedings of the 27th ACM International Conference on Information and Knowledge Management

Self Cite

View full text Add to dashboard Cite

show abstract

“…A common way to consider all factors is to build a comprehensive graph with multiple factors [16,15,40,25]. People have proposed many approaches to process such a comprehensive graph: [9,34] generate candidates routes from historical data and chooses one from candidates, which can choose routes according to the specific requirements of users.…”

Section: Route Recommendationmentioning

confidence: 99%

“…Some researchers [19,38,10] aims at recommending routes to taxis, to minimize congestion while maximizing profits. [15] utilizes transfer learning to alleviate insufficient data problem.…”

Section: Route Recommendationmentioning

confidence: 99%

Learning to Route via Theory-Guided Residual Network

Liu¹,

Zheng²,

Li³

2021

Preprint

View full text Add to dashboard Cite

The heavy traffic and related issues have always been concerns for modern cities. With the help of deep learning and reinforcement learning, people have proposed various policies to solve these traffic-related problems, such as smart traffic signal control systems and taxi dispatching systems. People usually validate these policies in a city simulator, since directly applying them in the real city introduces real cost. However, these policies validated in the city simulator may fail in the real city if the simulator is significantly different from the real world. To tackle this problem, we need to build a real-like traffic simulation system. Therefore, in this paper, we propose to learn the human routing model, which is one of the most essential part in the traffic simulator. This problem has two major challenges. First, human routing decisions are determined by multiple factors, besides the common time and distance factor. Second, current historical routes data usually covers just a small portion of vehicles, due to privacy and device availability issues. To address these problems, we propose a theory-guided residual network model, where the theoretical part can emphasize the general principles for human routing decisions (e.g., fastest route), and the residual part can capture drivable condition preferences (e.g., local road or highway). Since the theoretical part is composed of traditional shortest path algorithms that do not need data to train, our residual network can learn human routing models from limited data. We have conducted extensive experiments on multiple real-world datasets to show the superior performance of our model, especially with small data. Besides, we have also illustrated why our model is better at recovering real routes through case studies.

show abstract

“…Guo et al [16] diverged from the previous implementations by using a trajectory set for the actual path calculation. They generate a graph-like structure based on a user's historical trajectories as their routing infrastructure.…”

Section: Preference Mining Based On Historicalmentioning

confidence: 99%

SmarterROUTES—A Data-driven Context-aware Solution for Personalized Dynamic Routing and Navigation

Bock

Verstockt

2020

ACM Trans. Spatial Algorithms Syst.

View full text Add to dashboard Cite

SmarterROUTES contributes to personalised routing and navigation by data-driven route ranking and an environmentally aware road scene complexity-estimation mechanism. Traditional routing algorithms provide the fastest, shortest, or most ecological route by calculating the lowest-cost path from A to B using an underlying network of weighted connections. Our implementation goes a step further and ranks a set of such routes. The ranking is the result of additional weighing based on governmental data sets, extracts of the OpenStreetMap (OSM) database, or periodically adapted extracts of web services (e.g., Representational state transfer APIs). The selected data sets and their relative contributions to the overall ranking mechanism can be dynamically adapted by the end-user. Another major contribution toward a fully personalised navigation experience is the implementation of a road scene complexity scoring mechanism. Road complexity is estimated based on the combined input from geospatial data sources, traffic data, sensor analysis, and Street View-based complexity analysis. The latter input source uses Street View images as input for a Densenet Convolutional Neural Network (CNN), pre-trained on buildings using Bag-of-Words and Structure-from-motion features, outputting an image descriptor. The current version uses an adapted version of this network to predict road complexity scores. The predicted values correspond with the subjective complexity judgements of the end-users.

show abstract

Learning to Route with Sparse Trajectory Sets

Cited by 51 publications

References 36 publications

Correlated Time Series Forecasting using Multi-Task Deep Neural Networks

Correlated Time Series Forecasting using Multi-Task Deep Neural Networks

Learning to Route via Theory-Guided Residual Network

SmarterROUTES—A Data-driven Context-aware Solution for Personalized Dynamic Routing and Navigation

Contact Info

Product

Resources

About