Hi Prof,
Can we apply machine learning algorithm for mining (frequent, high utility etc.)?

Hi,

Actually, machine learning is quite broad and even pattern mining can be viewed as machine learning. But I think that you are referring perhaps to neural networks?

I know that some papers such as those by Dang Nguyen have used pattern mining in conjunction with some neural networks, if I remember. But I am not aware of papers that have used neural networks to find the patterns. But I guess it must be possible.

By the way, another type of machine learning is the evolutionary and nature-inspired algorithms such as genetic algorithms, particle swarm optimization, etc. These algorithms have been used for pattern mining. The idea is that these algorithms will find some approximate solutions to the pattern mining problems. Maybe they will miss some patterns but they will be quite fast. Several of these algorithms are implemented in SPMF for example.

Hope this helps

Thanks @Philippe for mentioning my works.

Yes, in our previous works I have combined patter mining with machine learning, namely
1. Frequent itemset + Neural network for transaction classification: [Trans2Vec: Learning Transaction Embedding via Items and Frequent Itemsets, PAKDD 2018]
2. Sequential pattern + Neural network for sequence classification and clustering: [Sqn2Vec: Learning Sequence Representation via Sequential Patterns with a Gap Constraint, ECML-PKDD 2018]
3. Frequent subgraph + Neural network for graph classification: [Learning Graph Representation via Frequent Subgraphs, SDM 2018]

Their source codes are shared on Github: https://github.com/nphdang

Thanks a lot both of you for valued and informative comments. Actually, I have a plan to train model for pattern mining over stream data especially.
Could you please add any comments regarding this idea?

I am not sure what has been done recently for pattern mining in streams. I did not read on this topic for a while.