CEIR: Concept-based Explainable Image Representation Learning

TL;DR

CEIR introduces a concept-based, explainable image representation learning method that leverages pretrained models and GPT-4 generated concepts to improve interpretability and clustering performance in an unsupervised manner.

Contribution

The paper proposes a novel concept-based framework combining CBM, CLIP, GPT-4, and VAE to enhance interpretability and performance of unsupervised image representations.

Findings

Achieves state-of-the-art clustering on CIFAR datasets

Provides human-understandable concept attributions

Enables open-world concept extraction without fine-tuning

Abstract

In modern machine learning, the trend of harnessing self-supervised learning to derive high-quality representations without label dependency has garnered significant attention. However, the absence of label information, coupled with the inherently high-dimensional nature, improves the difficulty for the interpretation of learned representations. Consequently, indirect evaluations become the popular metric for evaluating the quality of these features, leading to a biased validation of the learned representation rationale. To address these challenges, we introduce a novel approach termed Concept-based Explainable Image Representation (CEIR). Initially, using the Concept-based Model (CBM) incorporated with pretrained CLIP and concepts generated by GPT-4, we project input images into a concept vector space. Subsequently, a Variational Autoencoder (VAE) learns the latent representation from these projected concepts, which serves as the final image representation. Due to the capability of the representation to encapsulate high-level, semantically relevant concepts, the model allows for attributions to a human-comprehensible concept space. This not only enhances interpretability but also preserves the robustness essential for downstream tasks. For instance, our method exhibits state-of-the-art unsupervised clustering performance on benchmarks such as CIFAR10, CIFAR100, and STL10. Furthermore, capitalizing on the universality of human conceptual understanding, CEIR can seamlessly extract the related concept from open-world images without fine-tuning. This offers a fresh approach to automatic label generation and label manipulation.