Speaker-Invariant Affective Representation Learning via Adversarial Training | IEEE Conference Publication | IEEE Xplore