Consider a single layer neural network defined as (x;)=g(Wx+b) where xRd,WRmd,bRm and g is defined as g(z)=z+c where R is a constant, and cRm is a vector. For the output layer, we define conditional probability as p(yx;,v)=yexp{v(y)(x;)+y}exp{v(y)(x;)+y} Show that for any parameter value v(y)Rd and y for yY, there are parameter values v(y) and y such that for all x,y, p(yx;,v)=yexp{v(y)x+y}exp{v(y)x+y} Consider a single layer neural network defined as (x;)=g(Wx+b) where xRd,WRmd,bRm and g is defined as g(z)=z+c where R is a constant, and cRm is a vector. For the output layer, we define conditional probability as p(yx;,v)=yexp{v(y)(x;)+y}exp{v(y)(x;)+y} Show that for any parameter value v(y)Rd and y for yY, there are parameter values v(y) and y such that for all x,y, p(yx;,v)=yexp{v(y)x+y}exp{v(y)x+y}