Benchmarks, Algorithms, and Metrics for Hierarchical Disentanglement

TL;DR

This paper introduces benchmarks, algorithms, and metrics specifically designed for learning hierarchical representations in complex generative models, addressing limitations of existing methods that assume flat, independent factors.

Contribution

It presents novel benchmarks, algorithms, and metrics tailored for hierarchical and mixed discrete-continuous generative factors in representation learning.

Findings

New benchmarks for hierarchical disentanglement

Algorithms capable of capturing hierarchical structures

Metrics to evaluate hierarchical disentanglement

Abstract

In representation learning, there has been recent interest in developing algorithms to disentangle the ground-truth generative factors behind a dataset, and metrics to quantify how fully this occurs. However, these algorithms and metrics often assume that both representations and ground-truth factors are flat, continuous, and factorized, whereas many real-world generative processes involve rich hierarchical structure, mixtures of discrete and continuous variables with dependence between them, and even varying intrinsic dimensionality. In this work, we develop benchmarks, algorithms, and metrics for learning such hierarchical representations.