Dual Contrastive Learning for Semi-Supervised Fault Diagnosis Under Extremely Low Label Rate | IEEE Journals & Magazine | IEEE Xplore