songmin dai

weakly supervised semantic segmentation (2) contrastive learning (1) anomaly detection (1) model robustness (1) self-supervised learning (1) weakly supervised learning (1) adversarial training (1) instance segmentation (1) positive unlabeled learning (1) facial landmark detection (1) generalization error (1) black-box attack (1) generalization bound (1) diffusion model (1) deep neural network (1) loss function (1) consistency regularization (1) generative adversarial network (1) noisy label (1) mask generation (1)