A Survey of Zero-Shot Learning

By Wei Wang, Vincent Wenchen Zheng, Han Yu, Chunyan Miao. 2019

777 citations recorded by the source corpus.

Most machine-learning methods focus on classifying instances whose classes have already been seen in training. In practice, many applications require classifying instances whose classes have not been seen previously. Zero-shot learning is a powerful and promising learning paradigm, in which the classes covered by training instances and the classes we aim to classify are disjoint. In this paper, we provide a comprehensive survey of zero-shot learning. First of all, we provide an overview of zero-shot learning. According to the data utilized in model optimization, we classify zero-shot learning into three learning settings. Second, we describe different semantic spaces adopted in existing zero-shot learning works. Third, we categorize existing zero-shot learning methods and introduce representative methods under each category. Fourth, we discuss different applications of zero-shot learning. Finally, we highlight promising future research directions of zero-shot learning.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Representation & Self-Supervised Learning

Citations in the corpus

Rethinking the Role of Demonstrations: What Makes In-Context Learning Work? (2022, methodology): “We use these datasets because they (1) are true low-resource datasets with less than 10K training examples, (2) include well-studied benchmarks from GLUE (Wang et al., 2018) and Super-GLUE (Wang et al., 2019a), and (3) cover diverse domains including science, social media, ﬁnance, and more.”
Continual Learning for Robotics (2019, background): “The extreme case of zero-shot learning consists of the ability to detect new classes while being trained with a disjoint set of classes [166].”
A Review of Generalized Zero-Shot Learning Methods (2020, background): “Although several comprehensive reviews of ZSL models can be found in the literature [3], [23], [25], [26], none of them include an in-depth survey and analysis of GZSL.”
Zero-shot Text Classification via Reinforced Self-training (2020, background): “(Wang et al., 2019).”
Learning to Prompt for Open-Vocabulary Object Detection with Vision-Language Model (2022, background): “To alleviate the reliability of DNNs on big data and elaborate annotations, different paradigms such as semisupervised learning [34], few-shot learning [25, 27, 36], zero-shot learning [21,30,33], self-supervised learning [31], open-set learning [6, 22, 29] and advanced training strategies [17, 35] ”
Improved Few-Shot Visual Classification (2019, positive): “This inspired the field of few-shot learning [42, 43] which aims to computationally mimic human reasoning and learning from limited data.”
Empowering Things With Intelligence: A Survey of the Progress, Challenges, and Opportunities in Artificial Intelligence of Things (2020, background): “ZSL can be useful in some AIoT application scenarios.”
Continual Learning in Deep Networks: an Analysis of the Last Layer (2021, background): “Nevertheless, the successes in few-shot learning [25, 15, 51] or fast adaptation [7] show that when the feature space is good enough, only a small set of data is enough to identify a class.”
Deep Learning for Free-Hand Sketch: A Survey (2020, background): “This section will discuss the current bottlenecks of the free-hand sketch community, mainly considering from the perspective of deep learning.”
A survey on deep multimodal learning for computer vision: advances, trends, applications, and datasets (2021, background): “This is why zero-shot learning [130] takes place.”
Generalized Out-of-Distribution Detection: A Survey (2021, neutral): “Zero-shot Learning Zero-shot learning [98] has a similar goal of novelty discovery, but follows the training-testing scheme.”
Security and Privacy Challenges of Large Language Models: A Survey (2024, neutral): “Clear principles of using these LLMs and/or other AI tools in scientific explorations should be established to ensure transparency, fairness, and trustworthiness [211].”
Domain Adaptation for Medical Image Analysis: A Survey (2021, neutral): “Thus,researchondomaingeneralization[35], [146]–[148]andzero-shotlearning[149]–[151]willbewelcome inthefuture.”
Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, background): “Zero-shot learning [22], [23] aims at learning models from seen classes and classify samples whose categories are unseen”