/kernel_mean_embedding_distribution-regression

code for the KDD 2019 workshop paper https://arxiv.org/abs/1904.10583. Kernel mean embedding for distribution regression.

Primary LanguageJupyter Notebook

KDD 2019 workshop paper

This is a multiple instance regression algorithm using stacking-like approach and kernel mean embedding on the predictions.

I found in my previous work that the simple baseline Instance-MIR algorithm was performing (perhaps) surprisingly well on 2 of the 5 real world datasets I used. I am thus considering improvements in order to make the Instance-MIR even more competitive.