Skip to content

Instantly share code, notes, and snippets.

@yvanst

yvanst/org2beamer.org

Last active January 7, 2020 04:04
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save yvanst/075f074736b39dd6bfb1b8260469710f to your computer and use it in GitHub Desktop.
Save yvanst/075f074736b39dd6bfb1b8260469710f to your computer and use it in GitHub Desktop.
Download ZIP
org mode export to beamer
@yvanst
Copy link
Author

yvanst commented Jan 7, 2020 

* 深度学习是什么
** 神经网络
*** text                                                   :B_column:BMCOL:
:PROPERTIES:
:BEAMER_env: column
:BEAMER_opt: [t]
:BEAMER_col: 0.6
:END:
- 神经网络是一些带有权重和激活函数的节点的图
- 图中的节点组织成layer的形式, layer和layer之间的节点有连接
- 深度学习(深度神经网络)将layer堆叠在一起, 形成不同的结构
- 在如今海量的数据集, 强大的硬件, 新的训练算法的情况下, 可以训练出很大的神经网络

*** img                                                    :B_column:BMCOL:
:PROPERTIES:
:BEAMER_env: column
:BEAMER_opt: [t]
:BEAMER_col: 0.4
:END:

#+ATTR_LATEX: :width=\textwidth
[[file:network.png]]
** 深度学习
- 通常的机器学习算法在训练数据量继续增加时, 训练得到的模型运行效果不会不断变好
  (算法有能力上限), 深度学习没有这个限制
- 深度学习尤其擅长于处理非结构化数据, 比如图片(computer vision), 音频(speech
  processing), 文字(natural language)
- 常见的深度学习框架: TensorFlow, MXNet, Keras, PyTorch

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment