arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1906.03038cs.LGcs.CVstat.ML
收藏

一种具有对抗性领域自适应的零样本学习生成框架

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

IIT Kanpur(印度理工学院坎普尔分校);Microsoft Research, India(微软研究院印度分院);University of Toronto(多伦多大学)

Varun Khare, Divyat Mahajan, Homanga Bharadhwaj, Vinay Verma, Piyush Rai

摘要:

提出了一种基于领域自适应的零样本学习生成框架。我们的框架解决了零样本学习中可见和不可见类分布之间的域转移问题,并通过开发一个通过对抗性领域自适应训练的生成模型来最大限度地减少这种转移。我们的方法是基于端到端学习的类分布的可见类和不可见类。为了使模型能够学习看不见的类的类分布,我们根据类属性信息(对可见和不可见的类都可用)来参数化这些类分布。这提供了一种非常简单的方法来学习任何看不见的类的类分布,只给出其类属性信息,而没有标记的训练数据。用对抗性领域自适应训练该模型进一步提供了针对来自可见类和不可见类的数据之间的分布失配的鲁棒性。我们的方法也提供了一种新的方法来训练基于神经网络的分类器,以克服在零样本学习的hubness问题。通过一组全面的实验,我们证明了我们的模型在各种基准数据集上与各种最先进的零样本学习模型相比具有更高的精度。实验代码可在github.com/vkkhare/ZSL-ADA上获得

英文摘要:

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 these class distributions in terms of the class attribute information (which is available for both seen and unseen classes). This provides a very simple way to learn the class distribution of any unseen class, given only its class attribute information, and no labeled training data. Training this model with adversarial domain adaptation further provides robustness against the distribution mismatch between the data from seen and unseen classes. Our approach also provides a novel way for training neural net based classifiers to overcome the hubness problem in zero-shot learning. Through a comprehensive set of experiments, we show that our model yields superior accuracies as compared to various state-of-the-art zero shot learning models, on a variety of benchmark datasets. Code for the experiments is available at github.com/vkkhare/ZSL-ADA

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑