Trade-off between reconstruction loss and feature alignment for domain generalization

Open in new window