计算图上的微积分:反向传播(以及其他挖坑待补充内容)

内容来源博客(需科学上网):http://colah.github.io/

大神的油管教学视频:https://www.youtube.com/user/BrandonRohrer

计算图参考阅读:http://colah.github.io/posts/2015-08-Backprop/

 

其他待补充内容:

RNN和LSTM:

  • http://colah.github.io/posts/2015-08-Understanding-LSTMs/
  • https://distill.pub/2016/augmented-rnns/

卷积和神经网络:

  • 模块化的观点:http://colah.github.io/posts/2014-07-Conv-Nets-Modular/
  • 理解卷积:http://colah.github.io/posts/2014-07-Understanding-Convolutions/
  • 团体卷积:http://colah.github.io/posts/2014-12-Groups-Convolution/
  • 解卷积和棋盘工件:https://distill.pub/2016/deconv-checkerboard/

网络可视化

  • Visualizing MNIST:http://colah.github.io/posts/2014-10-Visualizing-MNIST/
  • 网络可视化:https://distill.pub/2017/feature-visualization/
  • 网络可解释性:https://distill.pub/2018/building-blocks/
  • 深入研究inception网络:https://ai.googleblog.com/2015/06/inceptionism-going-deeper-into-neural.html
  • 用神经网络进行手写的四个实验:https://distill.pub/2016/handwriting/
  • 可视化,深度学习与人:http://colah.github.io/posts/2015-01-Visualizing-Representations/

 

你可能感兴趣的:(深度学习)