Neural Networks

Model Representation I : article

Neuron in the brain

Neural networks were develpoed as simulating neurons or networks of neurons in the brain

• Neural networks are a set of algorithms

原本 7/8 號要寫完這篇~ 結果 顆顆顆~~ 哈哈哈

今天生日耶~~ 祝我生日快樂~

• Terminology:
  • Sigmoid (logistic) activation function
  • θ (theta) parameters —> in nerual networks fields called weight
    終於知道為啥之前學的 CS231 都用w來解釋了~

Neural Network

•  (j)
  a   =  "activation" of unit i in layer j
   i
  

•  (j)
  Θ  = matrix of weights controlling function mapping from layer j to layer j+1
  

• If network has s  units in layer j,
                  j
                                  (j)
  s     units in layer j+1, then Θ    will be of dimension s    * ( s  +  1 )
   j+1                                                      j+1      j
  

• EX:

    If layer 1 has 2 input nodes and layer 2 has 4 activation nodes. 
      
                  (1)
    Dimension of Θ   is going to be 4×3 
      
    where s  = 2  and  
           j
      
    s      = 4,  so 
     j + 1
      
    s        *  ( s  + 1 ) = 4 * 3
     j + 1         j