TY - GEN
T1 - Evaluation of distance measures for NMF-based face recognition
AU - Xue, Yun
AU - TONG, Chong Sze
AU - Zhang, Weipeng
N1 - Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2006
Y1 - 2006
N2 - Non-negative matrix factorization (NMF) is an unsupervised learning algorithm that can extract parts from visual data. The goal of this technique is to find intuitive basis such that training examples can be faithfully reconstructed using linear combination of basis images which are restricted to non-negative values. Thus NMF basis images can be understood as localized features that correspond better with intuitive notions of parts of images. However, there has not been any systematic study to identify suitable distance measure for data classification in this space defined by the NMF basis images. In this article we evaluate the performance of 17 distance measures (which include most of the standard distance measures used in face recognition, as well as a new non-negative vector similarity coefficient-based (NVSC) distance that we advocate for use in NMF-based pattern recognition) between feature vectors based on the result of the non-negative matrix factorization (NMF) algorithm for face recognition. Recognition experiments are performed using the MIT-CBCL database, CMU AMP Face Expression database and YaleB database. The experiments show, that our NVSC distance is consistently among the best measures with respect to different databases. Moreover, using this new distance, we almost always achieve better result than the L1, L2 and Mahalanobis distance which are often used in pattern recognition.
AB - Non-negative matrix factorization (NMF) is an unsupervised learning algorithm that can extract parts from visual data. The goal of this technique is to find intuitive basis such that training examples can be faithfully reconstructed using linear combination of basis images which are restricted to non-negative values. Thus NMF basis images can be understood as localized features that correspond better with intuitive notions of parts of images. However, there has not been any systematic study to identify suitable distance measure for data classification in this space defined by the NMF basis images. In this article we evaluate the performance of 17 distance measures (which include most of the standard distance measures used in face recognition, as well as a new non-negative vector similarity coefficient-based (NVSC) distance that we advocate for use in NMF-based pattern recognition) between feature vectors based on the result of the non-negative matrix factorization (NMF) algorithm for face recognition. Recognition experiments are performed using the MIT-CBCL database, CMU AMP Face Expression database and YaleB database. The experiments show, that our NVSC distance is consistently among the best measures with respect to different databases. Moreover, using this new distance, we almost always achieve better result than the L1, L2 and Mahalanobis distance which are often used in pattern recognition.
KW - Distance measures
KW - Face recognition
KW - Nonnegative matrix factorization
UR - http://www.scopus.com/inward/record.url?scp=38549116028&partnerID=8YFLogxK
U2 - 10.1109/ICCIAS.2006.294216
DO - 10.1109/ICCIAS.2006.294216
M3 - Conference proceeding
AN - SCOPUS:38549116028
SN - 1424406056
SN - 9781424406050
T3 - 2006 International Conference on Computational Intelligence and Security, ICCIAS 2006
SP - 651
EP - 656
BT - 2006 International Conference on Computational Intelligence and Security, ICCIAS 2006
PB - IEEE Computer Society
T2 - 2006 International Conference on Computational Intelligence and Security, ICCIAS 2006
Y2 - 3 October 2006 through 6 October 2006
ER -