Delayed reinforcement learning for closed-loop object recognition

Jing Peng, Bir Bhanu

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

1 Citation (Scopus)

Abstract

Object recognition is a multi-level process requiring a sequence of algorithms at low, intermediate and high levels. Generally, such systems are open loop with no feedback between levels and assuring their robustness is a key challenge in computer vision research. A robust closed-loop system based on delayed reinforcement learning is introduced in this paper. The parameters of a multi-level system employed for model-based object recognition are learned. The method improves recognition results over time by using the output at the highest level as feedback for the learning system. It has been experimentally validated by learning the parameters of image segmentation and feature extraction and thereby recognizing 2D objects. The approach systematically controls feedback in a multi-level vision system and provides a potential solution to a long-standing problem in the field of computer vision.

Original languageEnglish
Title of host publicationTrack D
Subtitle of host publicationParallel and Connectionist Systems
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages310-314
Number of pages5
Volume4
ISBN (Print)081867282X, 9780818672828
DOIs
StatePublished - 1 Jan 1996
Event13th International Conference on Pattern Recognition, ICPR 1996 - Vienna, Austria
Duration: 25 Aug 199629 Aug 1996

Other

Other13th International Conference on Pattern Recognition, ICPR 1996
CountryAustria
CityVienna
Period25/08/9629/08/96

Fingerprint

Object recognition
Reinforcement learning
Computer vision
Feedback
Open systems
Image segmentation
Closed loop systems
Feedback control
Learning systems
Feature extraction

Cite this

Peng, J., & Bhanu, B. (1996). Delayed reinforcement learning for closed-loop object recognition. In Track D: Parallel and Connectionist Systems (Vol. 4, pp. 310-314). [547436] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPR.1996.547436
Peng, Jing ; Bhanu, Bir. / Delayed reinforcement learning for closed-loop object recognition. Track D: Parallel and Connectionist Systems. Vol. 4 Institute of Electrical and Electronics Engineers Inc., 1996. pp. 310-314
@inproceedings{e4fdc3e4631b4ec19f6e731d8acda03b,
title = "Delayed reinforcement learning for closed-loop object recognition",
abstract = "Object recognition is a multi-level process requiring a sequence of algorithms at low, intermediate and high levels. Generally, such systems are open loop with no feedback between levels and assuring their robustness is a key challenge in computer vision research. A robust closed-loop system based on delayed reinforcement learning is introduced in this paper. The parameters of a multi-level system employed for model-based object recognition are learned. The method improves recognition results over time by using the output at the highest level as feedback for the learning system. It has been experimentally validated by learning the parameters of image segmentation and feature extraction and thereby recognizing 2D objects. The approach systematically controls feedback in a multi-level vision system and provides a potential solution to a long-standing problem in the field of computer vision.",
author = "Jing Peng and Bir Bhanu",
year = "1996",
month = "1",
day = "1",
doi = "10.1109/ICPR.1996.547436",
language = "English",
isbn = "081867282X",
volume = "4",
pages = "310--314",
booktitle = "Track D",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

Peng, J & Bhanu, B 1996, Delayed reinforcement learning for closed-loop object recognition. in Track D: Parallel and Connectionist Systems. vol. 4, 547436, Institute of Electrical and Electronics Engineers Inc., pp. 310-314, 13th International Conference on Pattern Recognition, ICPR 1996, Vienna, Austria, 25/08/96. https://doi.org/10.1109/ICPR.1996.547436

Delayed reinforcement learning for closed-loop object recognition. / Peng, Jing; Bhanu, Bir.

Track D: Parallel and Connectionist Systems. Vol. 4 Institute of Electrical and Electronics Engineers Inc., 1996. p. 310-314 547436.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Delayed reinforcement learning for closed-loop object recognition

AU - Peng, Jing

AU - Bhanu, Bir

PY - 1996/1/1

Y1 - 1996/1/1

N2 - Object recognition is a multi-level process requiring a sequence of algorithms at low, intermediate and high levels. Generally, such systems are open loop with no feedback between levels and assuring their robustness is a key challenge in computer vision research. A robust closed-loop system based on delayed reinforcement learning is introduced in this paper. The parameters of a multi-level system employed for model-based object recognition are learned. The method improves recognition results over time by using the output at the highest level as feedback for the learning system. It has been experimentally validated by learning the parameters of image segmentation and feature extraction and thereby recognizing 2D objects. The approach systematically controls feedback in a multi-level vision system and provides a potential solution to a long-standing problem in the field of computer vision.

AB - Object recognition is a multi-level process requiring a sequence of algorithms at low, intermediate and high levels. Generally, such systems are open loop with no feedback between levels and assuring their robustness is a key challenge in computer vision research. A robust closed-loop system based on delayed reinforcement learning is introduced in this paper. The parameters of a multi-level system employed for model-based object recognition are learned. The method improves recognition results over time by using the output at the highest level as feedback for the learning system. It has been experimentally validated by learning the parameters of image segmentation and feature extraction and thereby recognizing 2D objects. The approach systematically controls feedback in a multi-level vision system and provides a potential solution to a long-standing problem in the field of computer vision.

UR - http://www.scopus.com/inward/record.url?scp=84898770061&partnerID=8YFLogxK

U2 - 10.1109/ICPR.1996.547436

DO - 10.1109/ICPR.1996.547436

M3 - Conference contribution

SN - 081867282X

SN - 9780818672828

VL - 4

SP - 310

EP - 314

BT - Track D

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Peng J, Bhanu B. Delayed reinforcement learning for closed-loop object recognition. In Track D: Parallel and Connectionist Systems. Vol. 4. Institute of Electrical and Electronics Engineers Inc. 1996. p. 310-314. 547436 https://doi.org/10.1109/ICPR.1996.547436