Assuming w is Gaussian-distributed with mean Wmpandvariance{covariance matrix A-1, show that the

Assuming w is Gaussian-distributed with mean Wmpandvariance{covariance matrix A-1, show that the probability distribution of a(x) is

This means that the marginalized output is:

This is to be contrasted with the output of the most probablenet work. The integral of a sigmoid times a Gaussian can be approximated by