Beta-VAE Reproducibility: Challenges and Extensions
This addresses reproducibility challenges in unsupervised learning for researchers, but it is incremental as it builds on prior critiques.
The paper investigates the reproducibility of Beta-VAE, finding evidence of its lack of reproducibility and expanding experiments to more complex datasets, including implementing an FID scoring metric for qualitative analysis.
$β$-VAE is a follow-up technique to variational autoencoders that proposes special weighting of the KL divergence term in the VAE loss to obtain disentangled representations. Unsupervised learning is known to be brittle even on toy datasets and a meaningful, mathematically precise definition of disentanglement remains difficult to find. Here we investigate the original $β$-VAE paper and add evidence to the results previously obtained indicating its lack of reproducibility. We also further expand the experimentation of the models and include further more complex datasets in the analysis. We also implement an FID scoring metric for the $β$-VAE model and conclude a qualitative analysis of the results obtained. We end with a brief discussion on possible future investigations that can be conducted to add more robustness to the claims.