Concept Generalization in Visual Representation Learning

TL;DR

This paper introduces ImageNet-CoG, a benchmark that measures how well visual models generalize to unseen concepts based on semantic relationships, revealing the impact of architecture and training methods.

Contribution

It proposes a new semantic-aware benchmark for concept generalization and provides a large-scale analysis of various models' performance on it.

Findings

Semantic relationships influence generalization performance.

Model architecture and training data significantly affect concept transferability.

Transformer models outperform convolutional ones in generalization tasks.

Abstract

Measuring concept generalization, i.e., the extent to which models trained on a set of (seen) visual concepts can be leveraged to recognize a new set of (unseen) concepts, is a popular way of evaluating visual representations, especially in a self-supervised learning framework. Nonetheless, the choice of unseen concepts for such an evaluation is usually made arbitrarily, and independently from the seen concepts used to train representations, thus ignoring any semantic relationships between the two. In this paper, we argue that the semantic relationships between seen and unseen concepts affect generalization performance and propose ImageNet-CoG, a novel benchmark on the ImageNet-21K (IN-21K) dataset that enables measuring concept generalization in a principled way. Our benchmark leverages expert knowledge that comes from WordNet in order to define a sequence of unseen IN-21K concept sets that are semantically more and more distant from the ImageNet-1K (IN-1K) subset, a ubiquitous training set. This allows us to benchmark visual representations learned on IN-1K out-of-the box. We conduct a large-scale study encompassing 31 convolution and transformer-based models and show how different architectures, levels of supervision, regularization techniques and use of web data impact the concept generalization performance.