My own computer systems engineering degree didn't involve anything like this, but that was a few years ago now. It would certainly be ideal for post graduate research though, with the right supervisor.
From a hobby perspective, I recently came up with a problem for which this might be ideal.
When I'm listening to music, I'm often in the mood for some things but not others. Such mood isn't static and can't easily be represented by MP3 tags, so I've been mulling over the possibility of using some form of adaptive algorithm to use when tracks are skipped or played and what other temporally close skips or listens are performed.
For instance, if I'm driving, I might want keep me awake music, so might skip quiet or soporific music. When I get home and want something to relax to, I would switch to letting quiet tracks play and skipping energetic music.
Ideally playing or skipping a track should both update the weights associated with that track in relation to other tracks associated with that mood, and define a transition to a new state (if the current state is incompatible with
a track I've just allowed to play or skilled).
Adapt this into a crowd surfed algorithm, so that your default mood weightings come from the cloud and you could easily have a hobby project with the level of complexity of a CEP system.