Kun-Peng Ning

active learning (3) continual learning (1) catastrophic forgetting (1) sampling strategy (1) model robustness (1) distributed learning (1) query learning (1) gaussian mixture model (1) adversarial perturbation (1) low-rank adaptation (1) annotation cost (1) parameter space (1) asynchronous learning (1) label querying (1) safety vulnerability (1) harmful behavior (1) large language model (1) language model safety (1) parameter collision (1) task orthogonality (1)