Comparing robustness of pairwise and multiclass neural-network systems for face recognition

J. Uglov
,
Livija Jakaite
,
Vitaly Schetinin
,
Carsten Maple

University of Bedfordshire

Research Output:

Contribution to journal

Article

Peer-review

Open access

Abstract

Noise, corruptions, and variations in face images can seriously hurt the performance of face-recognition systems. To make these systems robust to noise and corruptions in image data, multiclass neural networks capable of learning from noisy data have been suggested. However on large face datasets such systems cannot provide the robustness at a high level. In this paper, we explore a pairwise neural-network system as an alternative approach to improve the robustness of face recognition. In our experiments, the pairwise recognition system is shown to outperform the multiclass-recognition system in terms of the predictive accuracy on the test face images.

Publication Information

Output type

Research Output:

Contribution to journal

Article

Peer-review

Original language

English

Article number

468693

Journal (Volume, Issue Number)

Eurasip Journal on Advances in Signal Processing (Volume 2008)

Publication milestones

Published - 06/12/2007

Publication status

Published - 06/12/2007

ISSN

1687-6172

Publication IDs

handle.net: 10547/293038
Scopus: 38749092949

Access to documents

10.1155/2008/468693

pairwise.pdf

Final published version, 838.84 KB

License:CC BY-NC-ND

https://asp-eurasipjournals.springeropen.com/articles/10.1155/2008/468693

Handle.net

Persistent link

Publication metrics

Metrics

Scopus

Download statistics

Download count

PlumX, opens in new tab

Captures