Yangdong Deng

backdoor attack (2) contrastive learning (1) zero-shot learning (1) adversarial learning (1) model selection (1) transfer learning (1) backdoor defense (1) generalization ability (1) adversarial attack (1) pre-trained model (1) black-box detection (1) inference latency (1) pretrained language model (1) model capacity (1) text similarity (1) concurrent computation (1) quasi-independent layer (1) llm-generated text detection (1) large language model (1) feature-based model selection (1)