<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[h_hhhhhworld的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/h_hhhhhworld</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; h_hhhhhworld]]></copyright><item><title><![CDATA[安装pytorch1.4.0方法]]></title><link>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/130575033</link><guid>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/130575033</guid><author>h_hhhhhworld</author><pubDate>Tue, 09 May 2023 10:40:00 +0800</pubDate><description><![CDATA[cpu，windows，三步。]]></description><category></category></item><item><title><![CDATA[【论文笔记】Details or Artifacts: A Locally Discriminative Learning Approach toRealistic Image Super-Reso]]></title><link>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/127507724</link><guid>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/127507724</guid><author>h_hhhhhworld</author><pubDate>Wed, 26 Oct 2022 16:13:13 +0800</pubDate><description><![CDATA[目前超分想要优异的纹理信息，那GAN（生成对抗网络 generative adversarial networks）就是必经之路。但GAN的训练是不稳定的，加入了GAN会生成丰富的细节，同事也会产生伪纹理（Artifacts），artifacts直接影响人的视觉感受。·本文中，作者开发了一个框架来区分GAN生成的伪纹理和真实细节，从而生成artifacts map以规范和稳定模型训练过程。]]></description><category></category></item><item><title><![CDATA[论文笔记：UA-MT]]></title><link>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/127324849</link><guid>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/127324849</guid><author>h_hhhhhworld</author><pubDate>Fri, 14 Oct 2022 21:20:19 +0800</pubDate><description><![CDATA[有标签数据和无标签数据都会输入到student和teacher model中，教师模型和学生模型的架构完全一致（神经网络层数之类的），学生模型利用lable data和预测值计算交叉熵L1，教师模型的预测值和学生模型的预测值计算一致性损失L2，最后L1+L2梯度下降更新学生模型的参数θ，教师模型的参数通过EMA的方式来更新。因此，认知不确定性测量的是训练过程本身所估计的模型参数的不确定性，这种不确定性是可以通过有针对性的调整（增加训练数据等方式）来缓解甚至是解决的。最终可以构成一个不确定性张量。]]></description><category></category></item><item><title><![CDATA[Anaconda安装|Pytorch安装|PyCharm安装|PyCharm配置torch]]></title><link>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/125534082</link><guid>https://blog.youkuaiyun.com/h_hhhhhworld/article/details/125534082</guid><author>h_hhhhhworld</author><pubDate>Thu, 30 Jun 2022 10:56:22 +0800</pubDate><description><![CDATA[Pytorch安装保姆教程]]></description><category></category></item></channel></rss>