TY - JOUR
T1 - Learning low-dimensional temporal representations with latent alignments
AU - Su, Bing
AU - Wu, Ying
N1 - Funding Information:
The authors would like to thank the anonymous reviewers for their valuable comments. This work was supported in part by the National Natural Science Foundation of China under Grant No.61603373, Youth Innovation Promotion Association CAS No. 2019110, National Science Foundation grant IIS-1619078, IIS-1815561, and the Army Research Office ARO W911NF-16-1-0138.
Publisher Copyright:
© 1979-2012 IEEE.
PY - 2020/11/1
Y1 - 2020/11/1
N2 - Low-dimensional discriminative representations enhance machine learning methods in both performance and complexity. This has motivated supervised dimensionality reduction (DR), which transforms high-dimensional data into a discriminative subspace. Most DR methods require data to be i.i.d. However, in some domains, data naturally appear in sequences, where the observations are temporally correlated. We propose a DR method, namely, latent temporal linear discriminant analysis (LT-LDA), to learn low-dimensional temporal representations. We construct the separability among sequence classes by lifting the holistic temporal structures, which are established based on temporal alignments and may change in different subspaces. We jointly learn the subspace and the associated latent alignments by optimizing an objective that favors easily separable temporal structures. We show that this objective is connected to the inference of alignments and thus allows for an iterative solution. We provide both theoretical insight and empirical evaluations on several real-world sequence datasets to show the applicability of our method.
AB - Low-dimensional discriminative representations enhance machine learning methods in both performance and complexity. This has motivated supervised dimensionality reduction (DR), which transforms high-dimensional data into a discriminative subspace. Most DR methods require data to be i.i.d. However, in some domains, data naturally appear in sequences, where the observations are temporally correlated. We propose a DR method, namely, latent temporal linear discriminant analysis (LT-LDA), to learn low-dimensional temporal representations. We construct the separability among sequence classes by lifting the holistic temporal structures, which are established based on temporal alignments and may change in different subspaces. We jointly learn the subspace and the associated latent alignments by optimizing an objective that favors easily separable temporal structures. We show that this objective is connected to the inference of alignments and thus allows for an iterative solution. We provide both theoretical insight and empirical evaluations on several real-world sequence datasets to show the applicability of our method.
KW - Dimensionality reduction
KW - discriminant analysis
KW - latent alignment
KW - temporal sequences
UR - http://www.scopus.com/inward/record.url?scp=85092454105&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85092454105&partnerID=8YFLogxK
U2 - 10.1109/TPAMI.2019.2919303
DO - 10.1109/TPAMI.2019.2919303
M3 - Article
C2 - 31144626
AN - SCOPUS:85092454105
SN - 0162-8828
VL - 42
SP - 2842
EP - 2857
JO - IEEE Transactions on Pattern Analysis and Machine Intelligence
JF - IEEE Transactions on Pattern Analysis and Machine Intelligence
IS - 11
M1 - 8723170
ER -