Code for hierarchical class clustering
Tianqi-py opened this issue · comments
Hi,
thanks for making the code publicly accessible. I was wondering if the code for creating the sequential multi-label datasets using the hierarchical class clustering algorithm is also published somewhere?
Thanks for your help!
Hi Tianqi,
Firstly, apologies for the late response.
As the data ids for both COCOseq and NUS-WIDEseq are released in the split state for benchmarking purposes, we did not release the code for creating the sequential multi-label datasets.
In the supplementary's Section 8.2 and Algorithm 2, we believe sufficient details were given about this topic (using sklearn's clustering algorithms should make your life even simpler).
Sincerely,
Chris.