PE1-6.2 Knowledge Representation
You can construct a Bayesian network and perform inference on it, apply hidden Markov models, and compare semantic networks, extended semantic networks and frames as representation schemes.
A Bayesian network is a compact way of writing a joint distribution by declaring what is conditionally independent of what, and that structure is what makes inference tractable at all. Hidden Markov models are the temporal case and are the reason this unit sits in an ECE course: HMMs run speech recognition, and they reappear in Unit 5. The semantic network and frame material is older symbolic AI, worth knowing as the tradition that preceded the probabilistic turn rather than as current practice.
Work through these
Bayesian network representation
A compact way to write down what depends on what, when everything is uncertain. The graph is the point: it says which independences you are assuming.
Construction of a Bayesian network
Building one from knowledge of a domain, including choosing the variable ordering. A poor ordering produces a correct but useless network, which is worth seeing.
Inference in Bayesian networks
Answering questions with one, exactly or approximately. Exact inference is intractable in general, which is why the approximate methods exist.
Hidden Markov models
The version of the above where time is the structure. It reappears in the speech application at the end of this course, so it is worth solid attention.
Approaches to knowledge representation
The wider question of how knowledge can be written down at all, before any specific scheme. It frames the three schemes below.
Knowledge representation using semantic networks
Representing knowledge as a graph of concepts and relations. Intuitive, and its weakness is that the meaning of a link is never quite pinned down.
Extended semantic networks
The refinements that address exactly that weakness. Comparing before and after is the content.
Knowledge representation using frames
Representing a concept as a bundle of slots with defaults. It is the ancestor of the object idea in programming, which is a useful connection to make.
Sign in to keep your progress.
Free resources
We haven't checked most of these for screen reader use yet.
Links last checked 30 Aug 2026.
Stuck here?
Ask a mentor. A real person answers, and they can see exactly which topic you're on. Usually within a couple of working days.
Checking your session…
Topics shown in module order.