Perceptron

• $f(x)=sign({\Sigma }_i{w}_i{x}_i+b)=sign({\mathbf{w}}^{\mathbf{T}}\mathbf{x}+b)$
  • $sign(x)=\{\begin{array}{ll}1& x\ge 0\\ 0& otherwise\end{array}$
• computational graph
• Multi layer
  • Stack multiple perceptrons
  • \begin{align} \\& h_0 = x h1= sign(\mathbf{w_1^T}+b_1) \\ &…\\& h1= sign(\mathbf{w_{L-1}^T}+b_L) \end{align}