Transfer Learning for Driving Pattern Recognition

Maoying Li, Liu Yang, Qinghua Hu, Chenyang Shen, Zhibin Du

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

Abstract

Driving pattern recognition based on driving status features (GPS, gear, and speed etc.) is of central importance in the development of intelligent transportation. While it is expensive and labor intensive to obtain a large amount of labeled driving data in real applications. It makes the driving pattern recognition particularly difficult for those domains without labeled data. In this paper, to tackle this challenging recognition task, we propose a novel and robust Transfer Learning method for Driving Pattern Recognition (TLDPR) that can transfer knowledge from other related source domains with labeled data to the target domain. Compared to the traditional supervised learning, one of the major difficulties of transfer learning is that the data from different domains may have distinct distributions. The proposed TLDPR is able to reduce the distribution difference in RKHS between the samples in target and source domain with the same driving pattern, and it can preserve the local manifold structure simultaneously. In addition, an iterative ensemble strategy is implemented to make the model more robust using the pseudo-labels. To evaluate the performance of TLDPR, comprehensive experiments have been conducted on parking lots datasets. The results show TLDPR can substantially outperform the state-of-the-art methods.

Original languageEnglish (US)
Title of host publicationPRICAI 2019
Subtitle of host publicationTrends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings
EditorsAbhaya C. Nayak, Alok Sharma
PublisherSpringer Verlag
Pages52-65
Number of pages14
ISBN (Print)9783030299101
DOIs
StatePublished - Jan 1 2019
Event16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019 - Yanuka Island, Fiji
Duration: Aug 26 2019Aug 30 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11671 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019
CountryFiji
CityYanuka Island
Period8/26/198/30/19

Fingerprint

Transfer Learning
Pattern Recognition
Pattern recognition
Knowledge Transfer
Target
Parking
Supervised learning
Supervised Learning
Gears
Global positioning system
Labels
Ensemble
Personnel
Distinct
Evaluate
Experiment

Keywords

  • Driving pattern
  • Maximum mean discrepancy
  • Transfer learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Li, M., Yang, L., Hu, Q., Shen, C., & Du, Z. (2019). Transfer Learning for Driving Pattern Recognition. In A. C. Nayak, & A. Sharma (Eds.), PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings (pp. 52-65). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11671 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-29911-8_5

Transfer Learning for Driving Pattern Recognition. / Li, Maoying; Yang, Liu; Hu, Qinghua; Shen, Chenyang; Du, Zhibin.

PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. ed. / Abhaya C. Nayak; Alok Sharma. Springer Verlag, 2019. p. 52-65 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11671 LNAI).

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

Li, M, Yang, L, Hu, Q, Shen, C & Du, Z 2019, Transfer Learning for Driving Pattern Recognition. in AC Nayak & A Sharma (eds), PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11671 LNAI, Springer Verlag, pp. 52-65, 16th Pacific Rim International Conference on Artificial Intelligence, PRICAI 2019, Yanuka Island, Fiji, 8/26/19. https://doi.org/10.1007/978-3-030-29911-8_5
Li M, Yang L, Hu Q, Shen C, Du Z. Transfer Learning for Driving Pattern Recognition. In Nayak AC, Sharma A, editors, PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. Springer Verlag. 2019. p. 52-65. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-29911-8_5
Li, Maoying ; Yang, Liu ; Hu, Qinghua ; Shen, Chenyang ; Du, Zhibin. / Transfer Learning for Driving Pattern Recognition. PRICAI 2019: Trends in Artificial Intelligence - 16th Pacific Rim International Conference on Artificial Intelligence, Proceedings. editor / Abhaya C. Nayak ; Alok Sharma. Springer Verlag, 2019. pp. 52-65 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{c32e3ab18ae44b6c87e80e5276a99b56,
title = "Transfer Learning for Driving Pattern Recognition",
abstract = "Driving pattern recognition based on driving status features (GPS, gear, and speed etc.) is of central importance in the development of intelligent transportation. While it is expensive and labor intensive to obtain a large amount of labeled driving data in real applications. It makes the driving pattern recognition particularly difficult for those domains without labeled data. In this paper, to tackle this challenging recognition task, we propose a novel and robust Transfer Learning method for Driving Pattern Recognition (TLDPR) that can transfer knowledge from other related source domains with labeled data to the target domain. Compared to the traditional supervised learning, one of the major difficulties of transfer learning is that the data from different domains may have distinct distributions. The proposed TLDPR is able to reduce the distribution difference in RKHS between the samples in target and source domain with the same driving pattern, and it can preserve the local manifold structure simultaneously. In addition, an iterative ensemble strategy is implemented to make the model more robust using the pseudo-labels. To evaluate the performance of TLDPR, comprehensive experiments have been conducted on parking lots datasets. The results show TLDPR can substantially outperform the state-of-the-art methods.",
keywords = "Driving pattern, Maximum mean discrepancy, Transfer learning",
author = "Maoying Li and Liu Yang and Qinghua Hu and Chenyang Shen and Zhibin Du",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/978-3-030-29911-8_5",
language = "English (US)",
isbn = "9783030299101",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "52--65",
editor = "Nayak, {Abhaya C.} and Alok Sharma",
booktitle = "PRICAI 2019",

}

TY - GEN

T1 - Transfer Learning for Driving Pattern Recognition

AU - Li, Maoying

AU - Yang, Liu

AU - Hu, Qinghua

AU - Shen, Chenyang

AU - Du, Zhibin

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Driving pattern recognition based on driving status features (GPS, gear, and speed etc.) is of central importance in the development of intelligent transportation. While it is expensive and labor intensive to obtain a large amount of labeled driving data in real applications. It makes the driving pattern recognition particularly difficult for those domains without labeled data. In this paper, to tackle this challenging recognition task, we propose a novel and robust Transfer Learning method for Driving Pattern Recognition (TLDPR) that can transfer knowledge from other related source domains with labeled data to the target domain. Compared to the traditional supervised learning, one of the major difficulties of transfer learning is that the data from different domains may have distinct distributions. The proposed TLDPR is able to reduce the distribution difference in RKHS between the samples in target and source domain with the same driving pattern, and it can preserve the local manifold structure simultaneously. In addition, an iterative ensemble strategy is implemented to make the model more robust using the pseudo-labels. To evaluate the performance of TLDPR, comprehensive experiments have been conducted on parking lots datasets. The results show TLDPR can substantially outperform the state-of-the-art methods.

AB - Driving pattern recognition based on driving status features (GPS, gear, and speed etc.) is of central importance in the development of intelligent transportation. While it is expensive and labor intensive to obtain a large amount of labeled driving data in real applications. It makes the driving pattern recognition particularly difficult for those domains without labeled data. In this paper, to tackle this challenging recognition task, we propose a novel and robust Transfer Learning method for Driving Pattern Recognition (TLDPR) that can transfer knowledge from other related source domains with labeled data to the target domain. Compared to the traditional supervised learning, one of the major difficulties of transfer learning is that the data from different domains may have distinct distributions. The proposed TLDPR is able to reduce the distribution difference in RKHS between the samples in target and source domain with the same driving pattern, and it can preserve the local manifold structure simultaneously. In addition, an iterative ensemble strategy is implemented to make the model more robust using the pseudo-labels. To evaluate the performance of TLDPR, comprehensive experiments have been conducted on parking lots datasets. The results show TLDPR can substantially outperform the state-of-the-art methods.

KW - Driving pattern

KW - Maximum mean discrepancy

KW - Transfer learning

UR - http://www.scopus.com/inward/record.url?scp=85072854052&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072854052&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-29911-8_5

DO - 10.1007/978-3-030-29911-8_5

M3 - Conference contribution

AN - SCOPUS:85072854052

SN - 9783030299101

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 52

EP - 65

BT - PRICAI 2019

A2 - Nayak, Abhaya C.

A2 - Sharma, Alok

PB - Springer Verlag

ER -