+1 Recommend
1 collections
      • Record: found
      • Abstract: found
      • Article: found

      Geospatial Data to Images: A Deep-Learning Framework for Traffic Forecasting


      Read this article at

          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.


          Traffic forecasting has been an active research field in recent decades, and with the development of deep-learning technologies, researchers are trying to utilize deep learning to achieve tremendous improvements in traffic forecasting, as it has been seen in other research areas, such as speech recognition and image classification. In this study, we summarize recent works in which deep-learning methods were applied for geospatial data-based traffic forecasting problems. Based on the insights from previous works, we further propose a deep-learning framework, which transforms geospatial data to images, and then utilizes the state-of-the-art deep-learning methodologies such as Convolutional Neural Network (CNN) and residual networks. To demonstrate the simplicity and effectiveness of our framework, we present a formulation of the New York taxi pick-up/drop-off forecasting problem, and show that our framework significantly outperforms traditional methods, including Historical Average (HA) and AutoRegressive Integrated Moving Average (ARIMA).

          Author and article information

          Tsinghua Science and Technology
          Tsinghua University Press (Xueyan Building, Tsinghua University, Beijing 100084, China )
          05 February 2019
          : 24
          : 1
          : 52-64
          [1]∙ Weiwei Jiang is with the Department of Electronic Engineering, Tsinghua University, Beijing 100084, China.
          [2]∙ Lin Zhang is with the Department of Electronic Engineering and Tsinghua-Berkeley Shenzhen Institute, Tsinghua University, Shenzhen 518055, China. E-mail: linzhang@ 123456tsinghua.edu.cn .
          Author notes
          * To whom correspondence should be addressed. E-mail: jww13@ 123456mails.tsinghua.edu.cn .

          Lin Zhang received the BSc, MSc, and PhD degrees from Tsinghua University, China, in 1998, 2001, and 2006, respectively. He is currently an associate professor in the Department of Electronic Engineering, Tsinghua University and an associate co-director with Tsinghua-Berkeley Shenzhen Institute. He is a recipient of IEEE/ACM IPSN Best Demo Award, 2014, IEEE CASE, Best Application Paper Award, 2013, etc. He has published more than 20 journal and conference papers. His research interests include efficient protocols for sensor networks, statistical learning and data mining algorithms for sensory data processing, and information theory.

          Weiwei Jiang received the BSc degree from Tsinghua University, China, in 2013. He is currently pursuing PhD degree in the same affiliation. He has published three conference papers before. His research interests include ridesharing service, incentive mechanism design, and behavior analysis in intelligent transportation system.

          Copyright @ 2019
          : 05 July 2017
          : 28 September 2017

          Software engineering,Data structures & Algorithms,Applied computer science,Computer science,Artificial intelligence,Hardware architecture
          geospatial data,deep learning,traffic forecasting,convolutional neural network,residual network


          Comment on this article