Abstract

Robust biometric recognition is of paramount importance in security and surveillance applications. In face based biometric systems, data is usually collected using a video camera with high frame rate and thus the captured data has high redundancy. Selecting the appropriate instances from this data to update a classification model, is a significant, yet valuable challenge. Active learning methods have gained popularity in identifying the salient and exemplar data instances from superfluous sets. Batch mode active learning schemes attempt to select a batch of samples simultaneously rather than updating the model after selecting every single data point. Existing work on batch mode active learning assume a fixed batch size, which is not a practical assumption in biometric recognition applications. In this paper, we propose a novel framework to dynamically select the batch size using clustering based unsupervised learning techniques. We also present a batch mode active learning strategy specially suited to handle the high redundancy in biometric datasets. The results obtained on the challenging VidTIMIT and MOBIO datasets corroborate the superiority of dynamic batch size selection over static batch size and also certify the potential of the proposed active learning scheme in being used for real world biometric recognition applications.

Original languageEnglish (US)
Title of host publicationProceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010
Pages15-22
Number of pages8
DOIs
StatePublished - 2010
Event9th International Conference on Machine Learning and Applications, ICMLA 2010 - Washington, DC, United States
Duration: Dec 12 2010Dec 14 2010

Other

Other9th International Conference on Machine Learning and Applications, ICMLA 2010
CountryUnited States
CityWashington, DC
Period12/12/1012/14/10

Fingerprint

Biometrics
Redundancy
Unsupervised learning
Video cameras
Problem-Based Learning

Keywords

  • Active learning
  • DBSCAN clustering
  • Numerical optimization

ASJC Scopus subject areas

  • Computer Science Applications
  • Human-Computer Interaction

Cite this

Chakraborty, S., Balasubramanian, V., & Panchanathan, S. (2010). Dynamic batch size selection for batch mode active learning in biometrics. In Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010 (pp. 15-22). [5708807] https://doi.org/10.1109/ICMLA.2010.10

Dynamic batch size selection for batch mode active learning in biometrics. / Chakraborty, Shayok; Balasubramanian, Vineeth; Panchanathan, Sethuraman.

Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010. 2010. p. 15-22 5708807.

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

Chakraborty, S, Balasubramanian, V & Panchanathan, S 2010, Dynamic batch size selection for batch mode active learning in biometrics. in Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010., 5708807, pp. 15-22, 9th International Conference on Machine Learning and Applications, ICMLA 2010, Washington, DC, United States, 12/12/10. https://doi.org/10.1109/ICMLA.2010.10
Chakraborty S, Balasubramanian V, Panchanathan S. Dynamic batch size selection for batch mode active learning in biometrics. In Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010. 2010. p. 15-22. 5708807 https://doi.org/10.1109/ICMLA.2010.10
Chakraborty, Shayok ; Balasubramanian, Vineeth ; Panchanathan, Sethuraman. / Dynamic batch size selection for batch mode active learning in biometrics. Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010. 2010. pp. 15-22
@inproceedings{b2ae7cb48ae3480eb4ffd9d7591a60d1,
title = "Dynamic batch size selection for batch mode active learning in biometrics",
abstract = "Robust biometric recognition is of paramount importance in security and surveillance applications. In face based biometric systems, data is usually collected using a video camera with high frame rate and thus the captured data has high redundancy. Selecting the appropriate instances from this data to update a classification model, is a significant, yet valuable challenge. Active learning methods have gained popularity in identifying the salient and exemplar data instances from superfluous sets. Batch mode active learning schemes attempt to select a batch of samples simultaneously rather than updating the model after selecting every single data point. Existing work on batch mode active learning assume a fixed batch size, which is not a practical assumption in biometric recognition applications. In this paper, we propose a novel framework to dynamically select the batch size using clustering based unsupervised learning techniques. We also present a batch mode active learning strategy specially suited to handle the high redundancy in biometric datasets. The results obtained on the challenging VidTIMIT and MOBIO datasets corroborate the superiority of dynamic batch size selection over static batch size and also certify the potential of the proposed active learning scheme in being used for real world biometric recognition applications.",
keywords = "Active learning, DBSCAN clustering, Numerical optimization",
author = "Shayok Chakraborty and Vineeth Balasubramanian and Sethuraman Panchanathan",
year = "2010",
doi = "10.1109/ICMLA.2010.10",
language = "English (US)",
isbn = "9780769543000",
pages = "15--22",
booktitle = "Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010",

}

TY - GEN

T1 - Dynamic batch size selection for batch mode active learning in biometrics

AU - Chakraborty, Shayok

AU - Balasubramanian, Vineeth

AU - Panchanathan, Sethuraman

PY - 2010

Y1 - 2010

N2 - Robust biometric recognition is of paramount importance in security and surveillance applications. In face based biometric systems, data is usually collected using a video camera with high frame rate and thus the captured data has high redundancy. Selecting the appropriate instances from this data to update a classification model, is a significant, yet valuable challenge. Active learning methods have gained popularity in identifying the salient and exemplar data instances from superfluous sets. Batch mode active learning schemes attempt to select a batch of samples simultaneously rather than updating the model after selecting every single data point. Existing work on batch mode active learning assume a fixed batch size, which is not a practical assumption in biometric recognition applications. In this paper, we propose a novel framework to dynamically select the batch size using clustering based unsupervised learning techniques. We also present a batch mode active learning strategy specially suited to handle the high redundancy in biometric datasets. The results obtained on the challenging VidTIMIT and MOBIO datasets corroborate the superiority of dynamic batch size selection over static batch size and also certify the potential of the proposed active learning scheme in being used for real world biometric recognition applications.

AB - Robust biometric recognition is of paramount importance in security and surveillance applications. In face based biometric systems, data is usually collected using a video camera with high frame rate and thus the captured data has high redundancy. Selecting the appropriate instances from this data to update a classification model, is a significant, yet valuable challenge. Active learning methods have gained popularity in identifying the salient and exemplar data instances from superfluous sets. Batch mode active learning schemes attempt to select a batch of samples simultaneously rather than updating the model after selecting every single data point. Existing work on batch mode active learning assume a fixed batch size, which is not a practical assumption in biometric recognition applications. In this paper, we propose a novel framework to dynamically select the batch size using clustering based unsupervised learning techniques. We also present a batch mode active learning strategy specially suited to handle the high redundancy in biometric datasets. The results obtained on the challenging VidTIMIT and MOBIO datasets corroborate the superiority of dynamic batch size selection over static batch size and also certify the potential of the proposed active learning scheme in being used for real world biometric recognition applications.

KW - Active learning

KW - DBSCAN clustering

KW - Numerical optimization

UR - http://www.scopus.com/inward/record.url?scp=79952411828&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952411828&partnerID=8YFLogxK

U2 - 10.1109/ICMLA.2010.10

DO - 10.1109/ICMLA.2010.10

M3 - Conference contribution

AN - SCOPUS:79952411828

SN - 9780769543000

SP - 15

EP - 22

BT - Proceedings - 9th International Conference on Machine Learning and Applications, ICMLA 2010

ER -