A Simple Exponential Family Framework for Zero-Shot Learning

By Vinay Kumar Verma, Piyush Rai. 2017

206 citations recorded by the source corpus.

We present a simple generative framework for learning to predict previously unseen classes, based on estimating class-attribute-gated class-conditional distributions. We model each class-conditional distribution as an exponential family distribution and the parameters of the distribution of each seen/unseen class are defined as functions of the respective observed class attributes. These functions can be learned using only the seen class data and can be used to predict the parameters of the class-conditional distribution of each unseen class. Unlike most existing methods for zero-shot learning that represent classes as fixed embeddings in some vector space, our generative model naturally represents each class as a probability distribution. It is simple to implement and also allows leveraging additional unlabeled data from unseen classes to improve the estimates of their class-conditional distributions using transductive/semi-supervised learning. Moreover, it extends seamlessly to few-shot learning by easily updating these distributions when provided with a small number of additional labelled examples from unseen classes. Through a comprehensive set of experiments on several benchmark data sets, we demonstrate the efficacy of our framework.

Subfields: Artificial Intelligence, Radiology, Nuclear Medicine and Imaging

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 Radiology, Nuclear Medicine and Imaging

Citations in the corpus

Dynamic VAEs with Generative Replay for Continual Zero-shot Learning (2021, background): “Generative approaches have received quite an attention over embedding approaches due to the ability to generate synthetic data of unseen classes[37, 38, 24] using only class attributes.”
Relational Generalized Few-Shot Learning (2019, background): “Alternatives to meta-learning include using pre-trained image features [23, 24] or learning a generative model that models the global class structure in the joint label space [25].”
Generalized Zero-Shot Learning via Synthesized Examples (2017, background): “Also in transductive setting EF-ZSL[30] has the 65.”
A Generative Model for Zero Shot Learning Using Conditional Variational Autoencoders (2017, background): “Hence, non linear compatibility learning methods have also been proposed[40, 42].”
F-VAEGAN-D2: A Feature Generating Framework for Any-Shot Learning (2019, background): “GFZSL [52] treats unknown labels of unseen class images as latent variables and applies Expectation-Maximization (EM).”
Normalization Matters in Zero-Shot Learning (2020, unknown): “9 GFZSL [45] 0.”
Zero-Shot Semantic Segmentation (2019, background): “There are different variants in the literature on how the projections or the similarity measure are computed: simple linear projection [1, 2, 6, 16, 24, 36], non-linear multi-modal embeddings [39, 43] or even hybrid methods [8, 32, 42, 48].”
Zero-shot Image Recognition Using Relational Matching, Adaptation and Calibration (2019, unknown): “8 GFZSL [32] 63.”
Generative Replay-based Continual Zero-Shot Learning (2021, methodology): “Then at testing time, predict a semantic vector for unseen class samples and assign a class based on the nearest neighbor al-gorithm [8, 6, 9, 10].”
A Zero-Shot Framework for Sketch-based Image Retrieval (2018, background): “Zero-shot learning in Image Classiﬁcation [25,26,27,28] refers to learning to recognize images of novel classes although no examples from these classes are present in the training set.”
Semantic Projection Network for Zero- and Few-Label Semantic Segmentation (2019, methodology): “Both ALE [1] and SJE [2] utilize the visual-semantic hinge loss, SYNC [7] align visual and semantic embedding space using manifold learning, and GFZSL [45] learns a generative model to capture the class conditional distribution.”
A Meta-Learning Framework for Generalized Zero-Shot Learning (2019, methodology): “This setting is popularly known as Zero-Shot Learning (ZSL), and has drawn a considerable interest recently [44, 32, 47, 5, 38, 53, 46, 27, 39, 7, 29, 4, 33, 10, 62, 41, 23].”
Zero-Shot Learning via Class-Conditioned Deep Generative Models (2017, methodology): “Some other recent works have explored the idea of generative models for zero-shot learning (Li et al. 2017; Verma et al. 2017).”
Small Sample Learning in Big Data Era (2018, background): “More generally, Verma and Rai (2017) modeled each classconditional distribution as an exponential family distribution and the parameters of the distribution of each seen/unseen class are defined as functions of the respective seen class attributes.”