Deep Learning Frameworks

• Today,
  there are many deep learning frameworks that makes it easy for you
  to implement neural networks.
  So rather than too strongly endorsing any of these frameworks
  i want to recommend you use to choose frameworks.
  (framwork를 사용하는 기준)

1. One important criteria is the ease of programming,
   and that means both developing the neural network and iterating on it
   as well as deploying it for production.
   개발하는 것 뿐만 아니라 production에 배포하는 것을 의미한다.
2. A second important criteria is running speeds.
3. And then, one criteria that people don't often talk about but i think is important is whether or not the framework is truly open.
   framework가 완전히 open되기 위해서는 open source 뿐만 아니라 good governance도 필요하다.
   아쉽게도 많은 회사들은 open source SW를 사용하면서도 단일 회사의 통제력을 유지했다.
   그래서 특정 framework가 open source로 얼마나 유지할 것인지, 그리고
   open soure가 한 기업의 통제 하에 어떤 이유에서든 미래에 바뀔 것인지에 대한 생각을 하고 있어야 한다.
   단기적으로는 programming language 선호도에 따라 선택해도 된다.

TensorFlow

• Cost Function $J(w) = w^2 - 10w + 25$라고 가정하자.
  우리는 Cost Function을 minimization하기 위해 $w=5$라는 것을 알고 있다.
  이를 TensorFlow를 통해서 구현해보자.
  

• Unless you have some training data $X, Y$,
  and you're training a neural network with a cost function depends on your data as well as parameters $w$.
  

The heart of the TensorFlow program is to compute to cost.
And then TensorFlow automatically figures out the derivatives and how to minimize the cost.
The nice thing about TensorFlow is that by implementing base the forward prop,
TensorFlow ill automatically figure out all the necessary backward steps needed to implement back-prop.