SINC: Self-Supervised In-Context Learning for Vision-Language Tasks | IEEE Conference Publication | IEEE Xplore