Sequence prediction tasks
sequence prediction tasks
Feedforward networks are fine for classifying objects, but their units (as distinct from their weights) have no memory of previous inputs. Consequently they are unable to cope with sequence prediction tasks - tasks like predicting, given a sequence of sunspot activity counts, what the sunspot activity for the next time period will be, and financial prediction tasks (e.g. given share prices for the last n days, and presumably other economic data, etc., predict tomorrow's share price).
Simple recurrent nets can tackle tasks like this, because they do have a kind of memory for recording information derived from activation values from previous time steps.
This article is included for general interest - sequence prediction tasks are not part of the syllabus of COMP9414 Artificial Intelligence.
Vous devez demander un compte pour contribuer à cette page.