A Generative Framework for Zero-Shot Learning with Adversarial Domain Adaptation

  title={A Generative Framework for Zero-Shot Learning with Adversarial Domain Adaptation},
  author={Varun Kumar Khare and Divyat Mahajan and Homanga Bharadhwaj and Vinay Kumar Verma and Piyush Rai},
  journal={2020 IEEE Winter Conference on Applications of Computer Vision (WACV)},
We present a domain adaptation based generative framework for zero-shot learning. Our framework addresses the problem of domain shift between the seen and unseen class distributions in zero-shot learning and minimizes the shift by developing a generative model trained via adversarial domain adaptation. Our approach is based on end-to-end learning of the class distributions of seen classes and unseen classes. To enable the model to learn the class distributions of unseen classes, we parameterize… 

Figures and Tables from this paper

Towards Zero-Shot Learning with Fewer Seen Class Examples

The proposed approach leverages meta-learning to train a deep generative model that integrates variational autoencoder and generative adversarial networks and outperforms state-of-the-art approaches by a significant margin when the number of examples per seen class is very small.

A Meta-Learning Framework for Generalized Zero-Shot Learning

This paper proposes a meta-learning based generative model based on integrating model-agnostic meta learning with a Wasserstein GAN (WGAN) to handle $(i)$ and $(ii)$, and uses a novel task distribution to handle ($ii)$.

Transductive Zero-Shot Learning using Cross-Modal CycleGAN

A new model for T-ZSL based upon CycleGAN jointly projects images on their seen class labels with a supervised objective and aligns unseen class labels and visual exemplars with adversarial and cycle-consistency objectives and shows the efficiency of the Cross-Modal CycleGAN model (CM-GAN) on the ImageNet T- ZSL task where it obtain state-of-the-art results.

Hardness Sampling for Self-Training Based Transductive Zero-Shot Learning

A more strict T-GZSL data setting is suggested and a new Self-Training framework with Hardness Sampling for T-ZSL is designed, called STHS, where an arbitrary inductive ZSL method could be seamlessly embedded and it is iteratively trained with unseen-class samples selected by the hardness sampling approach.

A Simple Approach for Zero-Shot Learning based on Triplet Distribution Embeddings

This work addresses the issue of expressivity in terms of modeling the intra-class variability for each class in Zero-Shot Learning by leveraging the use of distribution embeddings, which are modeled as Gaussian distributions.

VMAN: A Virtual Mainstay Alignment Network for Transductive Zero-Shot Learning

To align the weights for better knowledge transfer between seen/unseen classes, the virtual mainstay alignment network (VMAN) is proposed, which is tailored for the transductive ZSL task and modeled as a nested minimization problem and is solved by a Taylor approximate optimization paradigm.

An Iterative Co-Training Transductive Framework for Zero Shot Learning

An iterative co-training framework which contains two different base ZSL models and an exchanging module that can gradually boost the ZSL performance by fully exploiting the potential complementarity of the two models’ classification capabilities is introduced.

HardBoost: Boosting Zero-Shot Learning with Hard Classes

It is found that high semantic affinity among unseen classes is a plausible underlying cause of hardness and two metrics to detect hard classes are designed and designed.

ZERO - Detect objects without training examples by knowing their parts

This paper proposes ZERO: a zero-shot learning method which learns to recognize objects by their parts, and is tested on the example use case of bicycle recognition, for which it outperforms few-shot object detection techniques.

Zero-sample surface defect detection and classification based on semantic feedback neural network

An Ensemble Co-training algorithm is proposed, which adaptively reduces the prediction error in image tag embedding from multiple angles and improves model performance by constructing semantic embedded modules and feedback mechanisms.



Zero-Shot Learning via Class-Conditioned Deep Generative Models

A deep generative model for Zero-Shot Learning that represents each seen/unseen class using a class-specific latent-space distribution, conditioned on class attributes, which facilitates learning highly discriminative feature representations for the inputs.

Feature Generating Networks for Zero-Shot Learning

A novel generative adversarial network (GAN) that synthesizes CNN features conditioned on class-level semantic information, offering a shortcut directly from a semantic descriptor of a class to a class-conditional feature distribution.

Adversarial Discriminative Domain Adaptation

It is shown that ADDA is more effective yet considerably simpler than competing domain-adversarial methods, and the promise of the approach is demonstrated by exceeding state-of-the-art unsupervised adaptation results on standard domain adaptation tasks as well as a difficult cross-modality object classification task.

A Simple Exponential Family Framework for Zero-Shot Learning

A simple generative framework for learning to predict previously unseen classes, based on estimating class-attribute-gated class-conditional distributions, which extends seamlessly to few-shot learning by easily updating these distributions when provided with a small number of additional labelled examples from unseen classes.

A Generative Model for Zero Shot Learning Using Conditional Variational Autoencoders

This work tries to generate the samples from the given attributes, using a conditional variational autoencoder, and uses the generated samples for classification of the unseen classes in zero shot learning in Image Classification.

Synthesizing Samples for Zero-shot Learning

A novel approach is proposed which turns the ZSL problem into a conventional supervised learning problem by synthesizing samples for the unseen classes by estimating the probability distribution of an unseen class by using the knowledge from seen classes and the class attributes.

Zero-Shot Learning Through Cross-Modal Transfer

This work introduces a model that can recognize objects in images even if no training data is available for the object class, and uses novelty detection methods to differentiate unseen classes from seen classes.

Synthesized Classifiers for Zero-Shot Learning

This work introduces a set of "phantom" object classes whose coordinates live in both the semantic space and the model space and demonstrates superior accuracy of this approach over the state of the art on four benchmark datasets for zero-shot learning.

Transductive Unbiased Embedding for Zero-Shot Learning

This paper proposes a straightforward yet effective method named Quasi-Fully Supervised Learning (QFSL) to alleviate the bias problem in Zero-Shot Learning, which outperforms existing state-of-the-art approaches by a huge margin.

Learning Joint Feature Adaptation for Zero-Shot Recognition

A novel framework of learning data-dependent feature transforms for scoring similarity between an arbitrary pair of source and target data instances to account for the wide variability in target domain and designs a specific algorithm involving bilinear similarity functions and regularized least squares as penalties for feature displacement.