Asanapplication,wecombineWDA with low-dimensional clustering techniques, such as K-means, to perform subspaceclustering.

As these systems become ubiquitous, so does the need tocurtail their behavior.

Invariant Risk Minimization (IRM) is a recently proposed framework for outof-distribution (o.o.d) generalization.

Withthetwo-branch architecture, itnaturally incorporates pre-trained moving statistics in BN layers during fine-tuning, exploiting more priorknowledge ofpre-trained networks.