trying to understand how this is unsupervised learning/theory behind

[klai001]

Dear all
thank you for developing this method! its very useful indeed.
I would like to understand more about this method/ i.e. how does this consider as unsupervised learning when the user defines the background and target datasets e.g. healthy vs disease? and how does it differ from the purpose of e.g. PLSDA? where class labels/categorial covariates are also specified?