Evaluation of Latent Space Disentanglement in the Presence of Interdependent Attributes

Author name not available (Why is that?)

Publication date: 11 October 2021

Abstract: Controllable music generation with deep generative models has become increasingly reliant on disentanglement learning techniques. However, current disentanglement metrics, such as mutual information gap (MIG), are often inadequate and misleading when used for evaluating latent representations in the presence of interdependent semantic attributes often encountered in real-world music datasets. In this work, we propose a dependency-aware information metric as a drop-in replacement for MIG that accounts for the inherent relationship between semantic attributes.

Has companion code repository: https://github.com/karnwatcharasupat/dependency-aware-mi-metrics

This page was built for publication: Evaluation of Latent Space Disentanglement in the Presence of Interdependent Attributes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6380007)