Saved in:
Bibliographic Details
Main Authors: Kong, Lingjing, Chen, Guangyi, Huang, Biwei, Xing, Eric P., Chi, Yuejie, Zhang, Kun
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2406.00519
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917892739563520
author Kong, Lingjing
Chen, Guangyi
Huang, Biwei
Xing, Eric P.
Chi, Yuejie
Zhang, Kun
author_facet Kong, Lingjing
Chen, Guangyi
Huang, Biwei
Xing, Eric P.
Chi, Yuejie
Zhang, Kun
contents Learning concepts from natural high-dimensional data (e.g., images) holds potential in building human-aligned and interpretable machine learning models. Despite its encouraging prospect, formalization and theoretical insights into this crucial task are still lacking. In this work, we formalize concepts as discrete latent causal variables that are related via a hierarchical causal model that encodes different abstraction levels of concepts embedded in high-dimensional data (e.g., a dog breed and its eye shapes in natural images). We formulate conditions to facilitate the identification of the proposed causal model, which reveals when learning such concepts from unsupervised data is possible. Our conditions permit complex causal hierarchical structures beyond latent trees and multi-level directed acyclic graphs in prior work and can handle high-dimensional, continuous observed variables, which is well-suited for unstructured data modalities such as images. We substantiate our theoretical claims with synthetic data experiments. Further, we discuss our theory's implications for understanding the underlying mechanisms of latent diffusion models and provide corresponding empirical evidence for our theoretical insights.
format Preprint
id arxiv_https___arxiv_org_abs_2406_00519
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Learning Discrete Concepts in Latent Hierarchical Models
Kong, Lingjing
Chen, Guangyi
Huang, Biwei
Xing, Eric P.
Chi, Yuejie
Zhang, Kun
Machine Learning
Artificial Intelligence
Learning concepts from natural high-dimensional data (e.g., images) holds potential in building human-aligned and interpretable machine learning models. Despite its encouraging prospect, formalization and theoretical insights into this crucial task are still lacking. In this work, we formalize concepts as discrete latent causal variables that are related via a hierarchical causal model that encodes different abstraction levels of concepts embedded in high-dimensional data (e.g., a dog breed and its eye shapes in natural images). We formulate conditions to facilitate the identification of the proposed causal model, which reveals when learning such concepts from unsupervised data is possible. Our conditions permit complex causal hierarchical structures beyond latent trees and multi-level directed acyclic graphs in prior work and can handle high-dimensional, continuous observed variables, which is well-suited for unstructured data modalities such as images. We substantiate our theoretical claims with synthetic data experiments. Further, we discuss our theory's implications for understanding the underlying mechanisms of latent diffusion models and provide corresponding empirical evidence for our theoretical insights.
title Learning Discrete Concepts in Latent Hierarchical Models
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2406.00519