Résumé:
Pattern mining consists of finding interesting, useful and pertinent patterns (data structures)
that exist among large amount of data. These discovered patterns can be used as actionable
knowledge directly or they can be used by other data mining methods as an input. Itemsets
represent the most basic type of pattern and are the most treated in this field. In the real world,
the actual data is for the most part uncertain.
Indeed, we are interested in our work on this type of data, and as a result, our work consists of
providing an approach for extracting frequent itemsets from uncertain data using deep
reinforcement learning, which has had a lot of success in a variety of domains.
Keywords: frequent itemset mining, high utility itemset mining, uncertain data, reinforcement learning, deep learning.
