Concept Boundary Vectors

TL;DR

This paper introduces concept boundary vectors, a new method for capturing the semantic meaning of concepts in model latent spaces, enhancing interpretability of neural network representations.

Contribution

It proposes concept boundary vectors derived from concept boundaries in latent space, demonstrating their effectiveness over existing concept activation vectors.

Findings

Concept boundary vectors effectively capture semantic concepts.

They outperform concept activation vectors in interpretability tasks.

Empirical results validate the proposed method.

Abstract

Machine learning models are trained with relatively simple objectives, such as next token prediction. However, on deployment, they appear to capture a more fundamental representation of their input data. It is of interest to understand the nature of these representations to help interpret the model's outputs and to identify ways to improve the salience of these representations. Concept vectors are constructions aimed at attributing concepts in the input data to directions, represented by vectors, in the model's latent space. In this work, we introduce concept boundary vectors as a concept vector construction derived from the boundary between the latent representations of concepts. Empirically we demonstrate that concept boundary vectors capture a concept's semantic meaning, and we compare their effectiveness against concept activation vectors.