# Learning Compositional Representations for Few-Shot Recognition

## TL;DR

This paper introduces a regularization technique that enables deep learning models to learn compositional, part-based representations, improving few-shot recognition by requiring fewer examples for new categories.

## Contribution

It proposes a simple regularization method that disentangles feature representations into attribute-based subspaces using category-level annotations, bridging the gap between human and machine learning.

## Key findings

- Improved few-shot learning performance on CUB-200-2011, SUN397, and ImageNet datasets.
- Disentangled attribute-based representations facilitate rapid generalization.
- Fewer examples needed to learn classifiers for novel categories.

## Abstract

One of the key limitations of modern deep learning approaches lies in the amount of data required to train them. Humans, by contrast, can learn to recognize novel categories from just a few examples. Instrumental to this rapid learning ability is the compositional structure of concept representations in the human brain --- something that deep learning models are lacking. In this work, we make a step towards bridging this gap between human and machine learning by introducing a simple regularization technique that allows the learned representation to be decomposable into parts. Our method uses category-level attribute annotations to disentangle the feature space of a network into subspaces corresponding to the attributes. These attributes can be either purely visual, like object parts, or more abstract, like openness and symmetry. We demonstrate the value of compositional representations on three datasets: CUB-200-2011, SUN397, and ImageNet, and show that they require fewer examples to learn classifiers for novel categories.