We introduce D omain-specific M asks for G eneralization, a model for improving both in-
domain and out-of-domain generalization performance. For domain generalization, the goal …

The rise of deep neural networks has led to several breakthroughs for semantic
segmentation. In spite of this, a model trained on source domain often fails to work properly …

Data-driven machine learning (ML) is promoted as one potential technology to be used in
next-generation wireless systems. This led to a large body of research work that applies ML …

Existing bias mitigation methods for DNN models primarily work on learning debiased
encoders. This process not only requires a lot of instance-level annotations for sensitive …

Generalizing visual recognition models trained on a single distribution to unseen input
distributions (ie domains) requires making them robust to superfluous correlations in the …

As AI systems have obtained significant performance to be deployed widely in our daily lives
and human society, people both enjoy the benefits brought by these technologies and suffer …