Zero-Shot Learning -- The Good, the Bad and the Ugly

TL;DR

This paper critically analyzes the current state of zero-shot learning, introduces a unified benchmark for fair comparison, and evaluates existing methods to identify limitations and guide future research.

Contribution

It defines a new standardized benchmark for zero-shot learning, unifying evaluation protocols and data splits, and provides an in-depth comparison of state-of-the-art methods.

Findings

Established a unified benchmark for zero-shot learning evaluation.

Compared various state-of-the-art methods in standard and generalized settings.

Identified key limitations and challenges in current zero-shot learning approaches.

Abstract

Due to the importance of zero-shot learning, the number of proposed approaches has increased steadily recently. We argue that it is time to take a step back and to analyze the status quo of the area. The purpose of this paper is three-fold. First, given the fact that there is no agreed upon zero-shot learning benchmark, we first define a new benchmark by unifying both the evaluation protocols and data splits. This is an important contribution as published results are often not comparable and sometimes even flawed due to, e.g. pre-training on zero-shot test classes. Second, we compare and analyze a significant number of the state-of-the-art methods in depth, both in the classic zero-shot setting but also in the more realistic generalized zero-shot setting. Finally, we discuss limitations of the current status of the area which can be taken as a basis for advancing it.