Locality and compositionality in zero-shot learning

TL;DR

This paper investigates how locality and compositionality in learned representations affect zero-shot learning, emphasizing the importance of these properties for generalization without relying on pre-training on large datasets.

Contribution

It demonstrates the significance of locality and compositionality in ZSL representations and highlights the benefits of local-aware models for improved generalization.

Findings

Locality and compositionality are crucial for ZSL generalization.

Pre-training on large datasets is not necessary to observe these effects.

Local-aware models enhance zero-shot learning performance.

Abstract

In this work we study locality and compositionality in the context of learning representations for Zero Shot Learning (ZSL). In order to well-isolate the importance of these properties in learned representations, we impose the additional constraint that, differently from most recent work in ZSL, no pre-training on different datasets (e.g. ImageNet) is performed. The results of our experiments show how locality, in terms of small parts of the input, and compositionality, i.e. how well can the learned representations be expressed as a function of a smaller vocabulary, are both deeply related to generalization and motivate the focus on more local-aware models in future research directions for representation learning.