TY - JOUR
T1 - Deep semi-supervised metric learning via identification of manifold memberships
AU - Zhuang, Furen
AU - Moulin, Pierre
N1 - Funding Information:
FZ is supported by the Agency for Science, Technology and Research, Singapore. PM’s research is supported by the National Science Foundation, United States of America.
Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - Three of the key challenges in semi-supervised metric learning are the difficulty in sampling loss-producing triplets, the difficulty in locating similar data which are faraway from the anchor points, and the difficulty in making the model robust to noisy predicted pseudolabels. We propose a method which allows the use of class-representative anchors (proxies), and avoids the computational costs associated with triplet sampling. Our new semi-supervised metric learning method propagates labels along mutual nearest neighbor pairs, so that faraway similar data can be drawn to the anchors, while data which are not along these paths (and hence not on the same manifold as the anchors) can be pushed away from these anchors. By assessing the number of different labels which were propagated to the same point, we obtain an estimate of the probability that our prediction of the pseudolabel is accurate, and hence able to attenuate the effect of uncertain pseudolabels on our model by factoring in the confidence of these predictions. We show the superiority of our method over various state-of-the-art methods on four diverse public datasets.
AB - Three of the key challenges in semi-supervised metric learning are the difficulty in sampling loss-producing triplets, the difficulty in locating similar data which are faraway from the anchor points, and the difficulty in making the model robust to noisy predicted pseudolabels. We propose a method which allows the use of class-representative anchors (proxies), and avoids the computational costs associated with triplet sampling. Our new semi-supervised metric learning method propagates labels along mutual nearest neighbor pairs, so that faraway similar data can be drawn to the anchors, while data which are not along these paths (and hence not on the same manifold as the anchors) can be pushed away from these anchors. By assessing the number of different labels which were propagated to the same point, we obtain an estimate of the probability that our prediction of the pseudolabel is accurate, and hence able to attenuate the effect of uncertain pseudolabels on our model by factoring in the confidence of these predictions. We show the superiority of our method over various state-of-the-art methods on four diverse public datasets.
KW - Image
KW - Metric
KW - Retrieval
KW - Semi-supervised
KW - Zeroshot
UR - http://www.scopus.com/inward/record.url?scp=85115085230&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85115085230&partnerID=8YFLogxK
U2 - 10.1109/ICASSP39728.2021.9414447
DO - 10.1109/ICASSP39728.2021.9414447
M3 - Conference article
AN - SCOPUS:85115085230
SN - 1520-6149
VL - 2021-June
SP - 1755
EP - 1759
JO - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
JF - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
T2 - 2021 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2021
Y2 - 6 June 2021 through 11 June 2021
ER -