Current browse context:
cs.LG
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: On the Transfer of Disentangled Representations in Realistic Settings
(Submitted on 27 Oct 2020 (v1), last revised 11 Mar 2021 (this version, v2))
Abstract: Learning meaningful representations that disentangle the underlying structure of the data generating process is considered to be of key importance in machine learning. While disentangled representations were found to be useful for diverse tasks such as abstract reasoning and fair classification, their scalability and real-world impact remain questionable. We introduce a new high-resolution dataset with 1M simulated images and over 1,800 annotated real-world images of the same setup. In contrast to previous work, this new dataset exhibits correlations, a complex underlying structure, and allows to evaluate transfer to unseen simulated and real-world settings where the encoder i) remains in distribution or ii) is out of distribution. We propose new architectures in order to scale disentangled representation learning to realistic high-resolution settings and conduct a large-scale empirical study of disentangled representations on this dataset. We observe that disentanglement is a good predictor for out-of-distribution (OOD) task performance.
Submission history
From: Andrea Dittadi [view email][v1] Tue, 27 Oct 2020 16:15:24 GMT (12895kb,D)
[v2] Thu, 11 Mar 2021 11:43:10 GMT (13292kb,D)
Link back to: arXiv, form interface, contact.