Weakly Supervised Disentanglement with Guarantees

Abstract

Learning disentangled representations that correspond to factors of variation in real-world data is critical to interpretable and human-controllable machine learning. Recently, concerns about the viability of learning disentangled representations in a purely unsupervised manner has spurred a shift toward the incorporation of weak supervision. However, there is currently no formalism that identifies when and how weak supervision will guarantee disentanglement. To address this issue, we provide a theoretical framework to assist in analyzing the disentanglement guarantees (or lack thereof) conferred by weak supervision when coupled with learning algorithms based on distribution matching. We empirically verify the guarantees and limitations of several weak supervision methods (restricted labeling, match-pairing, and rank-pairing), demonstrating the predictive power and usefulness of our theoretical framework.

arXiv commentICLR 2020

References in corpus (12)

• Semi-Supervised Learning with Deep Generative Models
• Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations
• Deep Convolutional Inverse Graphics Network
• Disentangling by Factorising
• cGANs with Projection Discriminator
• Towards a Definition of Disentangled Representations
• Multi-Level Variational Autoencoder: Learning Disentangled Representations from Grouped Observations
• Learning Deep Disentangled Embeddings with the F-Statistic Loss
• Sample Efficient Adaptive Text-to-Speech
• Learning Latent Subspaces in Variational Autoencoders
• Demystifying Inter-Class Disentanglement
• The Incomplete Rosetta Stone Problem: Identifiability Results for Multi-View Nonlinear ICA

Cited by in corpus (25)

• Weakly-Supervised Disentanglement Without Compromises
• Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style
• Learning Causal Semantic Representation for Out-of-Distribution Prediction
• On Disentangled Representations Learned From Correlated Data
• Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding
• Desiderata for Representation Learning: A Causal Perspective
• An Identifiable Double VAE For Disentangled Representations
• Learning and controlling the source-filter representation of speech with a variational autoencoder
• On the Transfer of Disentangled Representations in Realistic Settings
• Evaluating the Disentanglement of Deep Generative Models through Manifold Topology
• CoDeGAN: Contrastive Disentanglement for Generative Adversarial Network
• Visual Representation Learning Does Not Generalize Strongly Within the Same Domain
• DynamicVAE: Decoupling Reconstruction Error and Disentangled Representation Learning
• Measuring the Biases and Effectiveness of Content-Style Disentanglement
• Weakly-Supervised Reinforcement Learning for Controllable Behavior
• DISSECT: Disentangled Simultaneous Explanations via Concept Traversals
• Video Reenactment as Inductive Bias for Content-Motion Disentanglement
• DCI-ES: An Extended Disentanglement Framework with Connections to Identifiability
• Learning disentangled representations via product manifold projection
• No Representation without Transformation
• Mixture Representation Learning with Coupled Autoencoders
• Towards Better Understanding of Disentangled Representations via Mutual Information
• Disentanglement based Active Learning
• Boxhead: A Dataset for Learning Hierarchical Representations
• DAReN: A Collaborative Approach Towards Reasoning And Disentangling