An Improved Semi-Supervised VAE for Learning Disentangled Representations
This work addresses a crucial challenge in representation learning for machine learning researchers, but it is incremental as it extends prior methods with a novel supervision technique.
The paper tackles the problem of learning interpretable and disentangled representations in semi-supervised settings by introducing a label replacement method, which significantly improves disentanglement with very limited supervision, as demonstrated through extensive experiments on synthetic and real datasets.
Learning interpretable and disentangled representations is a crucial yet challenging task in representation learning. In this work, we focus on semi-supervised disentanglement learning and extend work by Locatello et al. (2019) by introducing another source of supervision that we denote as label replacement. Specifically, during training, we replace the inferred representation associated with a data point with its ground-truth representation whenever it is available. Our extension is theoretically inspired by our proposed general framework of semi-supervised disentanglement learning in the context of VAEs which naturally motivates the supervised terms commonly used in existing semi-supervised VAEs (but not for disentanglement learning). Extensive experiments on synthetic and real datasets demonstrate both quantitatively and qualitatively the ability of our extension to significantly and consistently improve disentanglement with very limited supervision.