Learning Alignments from Latent Space Structures
Ieva Kazlauskaite, Carl Henrik Ek, Neill D. F. Campbell · The University of Bath Online Publications Store (The University of Bath) · 2016
In this paper we present a model that is capable of learning alignments between high-dimensional data by exploiting low-dimensional structures. Specifically, our method uses a Gaussian process latent variable model (GP-LVM) to learn alignments and latent representations simultaneously. The results show that our model performs alignment implicitly and improves the smoothness of the low dimensional representations.