TY - GEN
T1 - DeepJDOT
T2 - 15th European Conference on Computer Vision, ECCV 2018
AU - Damodaran, Bharath Bhushan
AU - Kellenberger, Benjamin
AU - Flamary, Rémi
AU - Tuia, Devis
AU - Courty, Nicolas
N1 - Publisher Copyright:
© 2018, Springer Nature Switzerland AG.
PY - 2018/1/1
Y1 - 2018/1/1
N2 - In computer vision, one is often confronted with problems of domain shifts, which occur when one applies a classifier trained on a source dataset to target data sharing similar characteristics (e.g. same classes), but also different latent data structures (e.g. different acquisition conditions). In such a situation, the model will perform poorly on the new data, since the classifier is specialized to recognize visual cues specific to the source domain. In this work we explore a solution, named DeepJDOT, to tackle this problem: through a measure of discrepancy on joint deep representations/labels based on optimal transport, we not only learn new data representations aligned between the source and target domain, but also simultaneously preserve the discriminative information used by the classifier. We applied DeepJDOT to a series of visual recognition tasks, where it compares favorably against state-of-the-art deep domain adaptation methods.
AB - In computer vision, one is often confronted with problems of domain shifts, which occur when one applies a classifier trained on a source dataset to target data sharing similar characteristics (e.g. same classes), but also different latent data structures (e.g. different acquisition conditions). In such a situation, the model will perform poorly on the new data, since the classifier is specialized to recognize visual cues specific to the source domain. In this work we explore a solution, named DeepJDOT, to tackle this problem: through a measure of discrepancy on joint deep representations/labels based on optimal transport, we not only learn new data representations aligned between the source and target domain, but also simultaneously preserve the discriminative information used by the classifier. We applied DeepJDOT to a series of visual recognition tasks, where it compares favorably against state-of-the-art deep domain adaptation methods.
KW - Deep domain adaptation
KW - Optimal transport
U2 - 10.1007/978-3-030-01225-0_28
DO - 10.1007/978-3-030-01225-0_28
M3 - Conference contribution
AN - SCOPUS:85055447170
SN - 9783030012243
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 467
EP - 483
BT - Computer Vision – ECCV 2018 - 15th European Conference, 2018, Proceedings
A2 - Ferrari, Vittorio
A2 - Sminchisescu, Cristian
A2 - Weiss, Yair
A2 - Hebert, Martial
PB - Springer Verlag
Y2 - 8 September 2018 through 14 September 2018
ER -