《计算机应用研究》|Application Research of Computers

基于可变损失和流形正则化的生成对抗网络

GAN based on variable loss and manifold regularization

免费全文下载 (已被下载 次)  
获取PDF全文
作者 丁赛赛,吕佳
机构 重庆师范大学 a.计算机与信息科学学院;b.重庆市数字农业服务工程技术研究中心,重庆 401331
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)12-018-3607-05
DOI 10.19734/j.issn.1001-3695.2019.09.0531
摘要 针对生成对抗网络中鉴别器在少量标记样本上的分类精度较差以及对流形局部扰动的鲁棒性不足的问题,提出一种基于可变损失和流形正则化的生成对抗网络算法。当标记样本较少时,该算法在鉴别器中利用可变损失代替原有对抗损失以解决训练前期分类性能较差的鉴别器对半监督分类任务的不利影响。此外,在鉴别器可变损失的基础上加入流形正则项,通过惩罚鉴别器在流形上分类决策的变化提高鉴别器对局部扰动的鲁棒性。以生成样本的质量和半监督的分类精度作为算法的评价标准,并在数据集SVHN和CIFAR-10上完成了数值实验。与其他半监督算法的对比结果表明,该算法在使用少量带标记数据的情况下能得到质量更高的生成样本和精度更高的分类结果。
关键词 生成对抗网络; 局部扰动; 可变损失; 流形正则化; 半监督
基金项目 重庆市自然科学基金资助项目(cstc2014jcyjA40011)
重庆师范大学科研项目(YKC19018)
本文URL http://www.arocmag.com/article/01-2020-12-018.html
英文标题 GAN based on variable loss and manifold regularization
作者英文名 Ding Saisai, Lyu Jia
机构英文名 a.College of Computer & Information Sciences,b.Chongqing Center of Engineering Technology Research on Digital Agriculture Service,Chongqing Normal University,Chongqing 401331,China
英文摘要 Aiming at the problem of the discriminator's poor classification accuracy on a small number of labeled samples and insufficient robustness to the local perturbation of manifolds in the generative adversarial network, this paper proposed a gene-rative adversarial network based on variable loss and manifold regularization. The algorithm used a variable loss instead of the original discriminator to solve the adverse effect of the poorly trained classifier on the semi-supervised classification task. In addition, on the basis of variable loss of discriminator, it added manifold regular terms to improve the robustness of discriminator to local disturbance by punishing the variation of classification decision of discriminator on manifold. Using the quality of the generated samples and the semi-supervised classification accuracy as the evaluation criteria of the algorithm, it performed numerical experiments on the dataset SVHN and CIFAR-10. Comparing with other semi-supervised algorithms, the results show that the proposed algorithm can obtain higher quality generated samples and higher precision classification results with a small amount of labeled data.
英文关键词 generative adversarial network(GAN); local disturbance; variable loss; manifold regularization; semi-supervised
参考文献 查看稿件参考文献
 
收稿日期 2019/9/4
修回日期 2019/10/21
页码 3607-3611
中图分类号 TP391.41
文献标志码 A