In Part I we discussed inference and showed that for certain models this is computationally tractable. However, for many models of interest, one cannot perform inference exactly and approximations are required.
In Part V we discuss approximate inference methods, beginning with sampling-based approaches. These are popular and well known in many branches of the mathematical sciences, having their origins in chemistry and physics. We also discuss alternative deterministic approximate inference methods which in some cases can have remarkably accurate performance.
It is important to bear in mind that no single algorithm is going to be best on all inference tasks. For this reason, we attempt throughout to explain the assumptions behind the techniques so that one may select an appropriate technique for the problem at hand.
Review the options below to login to check your access.
Log in with your Cambridge Higher Education account to check access.
If you believe you should have access to this content, please contact your institutional librarian or consult our FAQ page for further information about accessing our content.