Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Digging Deeper: Learning Multi-Level Concept Hierarchies

Oscar Hill, Mateo Espinosa Zarlenga, Mateja Jamnik

TL;DR

Multi-Level Concept Splitting (MLCS), which discovers multi-level concept hierarchies from only top-level supervision, and Deep-HiCEMs, an architecture that represents these discovered hierarchies and enables interventions at multiple levels of abstraction are presented.

Abstract

Although concept-based models promise interpretability by explaining predictions with human-understandable concepts, they typically rely on exhaustive annotations and treat concepts as flat and independent. To circumvent this, recent work has introduced Hierarchical Concept Embedding Models (HiCEMs) to explicitly model concept relationships, and Concept Splitting to discover sub-concepts using only coarse annotations. However, both HiCEMs and Concept Splitting are restricted to shallow hierarchies. We overcome this limitation with Multi-Level Concept Splitting (MLCS), which discovers multi-level concept hierarchies from only top-level supervision, and Deep-HiCEMs, an architecture that represents these discovered hierarchies and enables interventions at multiple levels of abstraction. Experiments across multiple datasets show that MLCS discovers human-interpretable concepts absent during training and that Deep-HiCEMs maintain high accuracy while supporting test-time concept interventions that can improve task performance.

Digging Deeper: Learning Multi-Level Concept Hierarchies

TL;DR

Multi-Level Concept Splitting (MLCS), which discovers multi-level concept hierarchies from only top-level supervision, and Deep-HiCEMs, an architecture that represents these discovered hierarchies and enables interventions at multiple levels of abstraction are presented.

Abstract

Although concept-based models promise interpretability by explaining predictions with human-understandable concepts, they typically rely on exhaustive annotations and treat concepts as flat and independent. To circumvent this, recent work has introduced Hierarchical Concept Embedding Models (HiCEMs) to explicitly model concept relationships, and Concept Splitting to discover sub-concepts using only coarse annotations. However, both HiCEMs and Concept Splitting are restricted to shallow hierarchies. We overcome this limitation with Multi-Level Concept Splitting (MLCS), which discovers multi-level concept hierarchies from only top-level supervision, and Deep-HiCEMs, an architecture that represents these discovered hierarchies and enables interventions at multiple levels of abstraction. Experiments across multiple datasets show that MLCS discovers human-interpretable concepts absent during training and that Deep-HiCEMs maintain high accuracy while supporting test-time concept interventions that can improve task performance.
Paper Structure (32 sections, 3 figures, 5 tables)

This paper contains 32 sections, 3 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Concept learning
  4. Concept Splitting and HiCEMs
  5. Multi-Level Concept Splitting
  6. Hierarchical Sparse Autoencoder
  7. Depth of the Hierarchy
  8. Deep Hierarchical CEMs
  9. Concept Structure
  10. Architecture
  11. Sub-concepts Modules
  12. Experiments
  13. PseudoKitchens-2
  14. Setup
  15. Datasets
  16. ...and 17 more sections

Figures (3)

  • Figure 1: Deep-HiCEM: as in a HiCEM, from a latent code $\mathbf{h}$, we learn two embeddings per concept ($\mathbf{\hat{c}_i^{+\prime}}$ and $\mathbf{\hat{c}_i^{-\prime}}$), which are then passed through sub-concepts modules, which produce new embeddings ($\mathbf{\hat{c}_i^{+}}$ and $\mathbf{\hat{c}_i^{-}}$) that include information about sub-concepts and their descendants.
  • Figure 2: Task accuracy as discovered concepts are intervened on. Intervening on discovered concepts improves task accuracy, with a few exceptions that would benefit from further investigation.
  • Figure 3: Change in task accuracy as provided concepts are intervened on. Provided concept interventions work just as well in Deep-HiCEMs as they do in HiCEMs.