Transformer-Based Self-Supervised Multimodal Representation Learning for Wearable Emotion Recognition | IEEE Journals & Magazine | IEEE Xplore