As described in the lecture, for single$-$layer neural networks with quadratic loss, we have the following objective function $R(b)=\frac{1}{2}{\displaystyle \underset{n=1}{\overset{N}{}}}(y^n-(w^T{x}^n+b){)}^2,$ where we made the bias parameter $b$ explicit. Which bias parameter optimizes the objective function assuming fixed $w?$

W$$hlen Sie eine Antwort:

a$.,b=\frac{2}{N}{\displaystyle \underset{n=1}{\overset{N}{}}}(y^n-(w^T{x}^n))$

b$.b=\frac{1}{N}{\displaystyle \underset{n=1}{\overset{N}{}}}(x^n-(w^T{y}^n))$

c$.b=\frac{1}{2N}{\displaystyle \underset{n=1}{\overset{N}{}}}(y^n-(w^T{x}^n))$

d$.b=\frac{1}{N}{\displaystyle \underset{n=1}{\overset{N}{}}}(y^n-(w^T{x}^n))$

e$.b={\displaystyle \underset{n=1}{\overset{N}{}}}(y^n-(w^T{x}^n))$