On the Latent Space of Wasserstein Auto-Encoders
arXiv:1802.03761v155 citations
Originality Incremental advance
AI Analysis
This work addresses representation learning challenges for machine learning practitioners, but it is incremental as it builds on existing WAE methods.
The paper investigates the impact of latent space dimensionality in Wasserstein auto-encoders, finding that random encoders outperform deterministic ones and showing promising results on a disentanglement benchmark.
We study the role of latent space dimensionality in Wasserstein auto-encoders (WAEs). Through experimentation on synthetic and real datasets, we argue that random encoders should be preferred over deterministic encoders. We highlight the potential of WAEs for representation learning with promising results on a benchmark disentanglement task.