Proposal and Evaluation of Prediction of Pavement Rutting Depth by Recurrent Neural Network

Tomoyuki Okuda, Kouyu Suzuki, Naohiko Kohtake

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper describes a method of predicting the rutting depth by introducing Adam and dropout into MLP of neural network and GRU of recurrent neural net that can handle time series data. We built a model to predict the current rutting depth from the past rutting 3 years ago. We compared RMSE with the multiple regression model (MLR), which is most frequently used as a regression problem for the time variation of rutting depth. As a result, RMSE decreased in the order of MLR, MLP, GRU. The difference between GRU and RMSE of MLR was about 10%.

Original languageEnglish
Title of host publicationProceedings - 2017 6th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2017
EditorsKiyota Hashimoto, Naoki Fukuta, Tokuro Matsuo, Sachio Hirokawa, Masao Mori, Masao Mori
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1053-1054
Number of pages2
ISBN (Electronic)9781538606216
DOIs
Publication statusPublished - 2017 Nov 15
Event6th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2017 - Hamamatsu, Shizuoka, Japan
Duration: 2017 Jul 9 → …

Publication series

NameProceedings - 2017 6th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2017

Other

Other6th IIAI International Congress on Advanced Applied Informatics, IIAI-AAI 2017
CountryJapan
CityHamamatsu, Shizuoka
Period17/7/9 → …

Keywords

  • Machine learning
  • Pavement condition survey
  • Recurrent neural network
  • Road management

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Computer Science Applications
  • Information Systems
  • Information Systems and Management

Fingerprint Dive into the research topics of 'Proposal and Evaluation of Prediction of Pavement Rutting Depth by Recurrent Neural Network'. Together they form a unique fingerprint.

Cite this