Figure 2 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Source publication
Autoencoders and their variants are among the most widely used models in representation learning and generative modeling. However, autoencoder-based models usually assume that the learned representations are i.i.d. and fail to capture the correlations between the data samples. To address this issue, we propose a novel Sparse Gaussian Process Bayesi...
Context in source publication
Context 1
... begin our empirical evaluation by considering the moving ball dataset proposed by Pearce (2019 Fig. 2 illustrates the performance of the considered methods in terms of root mean squared error (RMSE). The results show that our GP-BAE model performs much better than GP-VAE (Pearce, 2019) though both models use the same full GP priors. In addition, by treating inducing inputs and kernel hyper-parameters of sparse GPs in a Bayesian ...
Similar publications
Gaussian process state-space models (GPSSMs) provide a principled and flexible approach to modeling the dynamics of a latent state, which is observed at discrete-time points via a likelihood model. However, inference in GPSSMs is computationally and statistically challenging due to the large number of latent variables in the model and the strong te...