A novel fuzzy feature encoding approach for image classification

Umit L. Altintakan, Adnan Yazici

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

Abstract

Feature encoding is a crucial step in BOW image representation. The standard BOW model assigns each image feature to the nearest visual-word without making a distinction between the features that are assigned to the same words. This hard feature assignment leads to high quantization errors and degrades the learning capacity of the classifiers in image classification. We propose a fuzzy feature encoding approach to overcome the uncertainty problem in BOW through assigning each image feature to the visual-words with some membership degrees. We employ two classification techniques, Naive Bayesian and SVM, to evaluate the effect of the fuzzy assignment in image classification. Experiments conducted on image datasets show that fuzzy feature encoding significantly improves the classification accuracy.

Original languageEnglish
Title of host publication2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1134-1139
Number of pages6
ISBN (Electronic)9781509006250
DOIs
Publication statusPublished - Nov 7 2016
Event2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016 - Vancouver, Canada
Duration: Jul 24 2016Jul 29 2016

Publication series

Name2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016

Conference

Conference2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016
CountryCanada
CityVancouver
Period7/24/167/29/16

Fingerprint

Image classification
Image Classification
Encoding
Assignment
Image Representation
Classifiers
Assign
Standard Model
Quantization
Classifier
Uncertainty
Evaluate
Experiments
Experiment
Vision

ASJC Scopus subject areas

  • Control and Optimization
  • Logic
  • Modelling and Simulation

Cite this

Altintakan, U. L., & Yazici, A. (2016). A novel fuzzy feature encoding approach for image classification. In 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016 (pp. 1134-1139). [07737815] (2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/FUZZ-IEEE.2016.7737815

A novel fuzzy feature encoding approach for image classification. / Altintakan, Umit L.; Yazici, Adnan.

2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 1134-1139 07737815 (2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016).

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

Altintakan, UL & Yazici, A 2016, A novel fuzzy feature encoding approach for image classification. in 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016., 07737815, 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016, Institute of Electrical and Electronics Engineers Inc., pp. 1134-1139, 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016, Vancouver, Canada, 7/24/16. https://doi.org/10.1109/FUZZ-IEEE.2016.7737815
Altintakan UL, Yazici A. A novel fuzzy feature encoding approach for image classification. In 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 1134-1139. 07737815. (2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016). https://doi.org/10.1109/FUZZ-IEEE.2016.7737815
Altintakan, Umit L. ; Yazici, Adnan. / A novel fuzzy feature encoding approach for image classification. 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 1134-1139 (2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016).
@inproceedings{494ddd5fce3b4477b41daa32afb0f169,
title = "A novel fuzzy feature encoding approach for image classification",
abstract = "Feature encoding is a crucial step in BOW image representation. The standard BOW model assigns each image feature to the nearest visual-word without making a distinction between the features that are assigned to the same words. This hard feature assignment leads to high quantization errors and degrades the learning capacity of the classifiers in image classification. We propose a fuzzy feature encoding approach to overcome the uncertainty problem in BOW through assigning each image feature to the visual-words with some membership degrees. We employ two classification techniques, Naive Bayesian and SVM, to evaluate the effect of the fuzzy assignment in image classification. Experiments conducted on image datasets show that fuzzy feature encoding significantly improves the classification accuracy.",
author = "Altintakan, {Umit L.} and Adnan Yazici",
year = "2016",
month = "11",
day = "7",
doi = "10.1109/FUZZ-IEEE.2016.7737815",
language = "English",
series = "2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "1134--1139",
booktitle = "2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016",
address = "United States",

}

TY - GEN

T1 - A novel fuzzy feature encoding approach for image classification

AU - Altintakan, Umit L.

AU - Yazici, Adnan

PY - 2016/11/7

Y1 - 2016/11/7

N2 - Feature encoding is a crucial step in BOW image representation. The standard BOW model assigns each image feature to the nearest visual-word without making a distinction between the features that are assigned to the same words. This hard feature assignment leads to high quantization errors and degrades the learning capacity of the classifiers in image classification. We propose a fuzzy feature encoding approach to overcome the uncertainty problem in BOW through assigning each image feature to the visual-words with some membership degrees. We employ two classification techniques, Naive Bayesian and SVM, to evaluate the effect of the fuzzy assignment in image classification. Experiments conducted on image datasets show that fuzzy feature encoding significantly improves the classification accuracy.

AB - Feature encoding is a crucial step in BOW image representation. The standard BOW model assigns each image feature to the nearest visual-word without making a distinction between the features that are assigned to the same words. This hard feature assignment leads to high quantization errors and degrades the learning capacity of the classifiers in image classification. We propose a fuzzy feature encoding approach to overcome the uncertainty problem in BOW through assigning each image feature to the visual-words with some membership degrees. We employ two classification techniques, Naive Bayesian and SVM, to evaluate the effect of the fuzzy assignment in image classification. Experiments conducted on image datasets show that fuzzy feature encoding significantly improves the classification accuracy.

UR - http://www.scopus.com/inward/record.url?scp=85006710367&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85006710367&partnerID=8YFLogxK

U2 - 10.1109/FUZZ-IEEE.2016.7737815

DO - 10.1109/FUZZ-IEEE.2016.7737815

M3 - Conference contribution

T3 - 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016

SP - 1134

EP - 1139

BT - 2016 IEEE International Conference on Fuzzy Systems, FUZZ-IEEE 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -