Hi All,
I have released the source code of GE-FSG which learns embeddings (aka feature vectors) for graphs. This work was published at SDM 2018 conference.
https://github.com/nphdang/GE-FSGOur method combines neural document embedding model with frequent subgraph mining to learn graph embeddings which are useful in many machine learning tasks such as classification and clustering.
Our component "fsg_miner.exe" can be used as a standalone tool to mine frequent subgraphs. It runs fast since it is implemented in parallel.
We plan to release the source code of transaction embedding and sequence embedding learning in the near future.
If you have any trouble when running the code, please feel free to let me know.
Cheers,
Dang Nguyen