Timezone: »
Measuring the similarity between data points often requires domain knowledge, which can in parts be compensated by relying on unsupervised methods such as latent-variable models, where similarity/distance is estimated in a more compact latent space. Prevalent is the use of the Euclidean metric, which has the drawback of ignoring information about similarity of data stored in the decoder, as captured by the framework of Riemannian geometry. We propose an extension to the framework of variational auto-encoders allows learning flat latent manifolds, where the Euclidean metric is a proxy for the similarity between data points. This is achieved by defining the latent space as a Riemannian manifold and by regularising the metric tensor to be a scaled identity matrix. Additionally, we replace the compact prior typically used in variational auto-encoders with a recently presented, more expressive hierarchical one---and formulate the learning problem as a constrained optimisation problem. We evaluate our method on a range of data-sets, including a video-tracking benchmark, where the performance of our unsupervised approach nears that of state-of-the-art supervised approaches, while retaining the computational efficiency of straight-line-based approaches.
Author Information
Nutan Chen (Volkswagen AG)
Alexej Klushyn (ML Research, VW Group)
Francesco Ferroni (Argo AI)
Justin Bayer (Volkswagen Group)
Patrick van der Smagt (Volkswagen Group)
More from the Same Authors
-
2021 : Exploration via Empowerment Gain: Combining Novelty, Surprise and Learning Progress »
Philip Becker-Ehmck · Maximilian Karl · Jan Peters · Patrick van der Smagt -
2022 : Probabilistic Dalek - Emulator framework with probabilistic prediction for supernova tomography »
Wolfgang Kerzendorf · Nutan Chen · Patrick van der Smagt -
2022 : Local distance preserving autoencoders using continuous kNN graphs »
Nutan Chen · Patrick van der Smagt · Botond Cseke -
2019 Poster: Switching Linear Dynamics for Variational Bayes Filtering »
Philip Becker-Ehmck · Jan Peters · Patrick van der Smagt -
2019 Oral: Switching Linear Dynamics for Variational Bayes Filtering »
Philip Becker-Ehmck · Jan Peters · Patrick van der Smagt