周次

研究方向

大模型安全-文本对抗

本周完成工作

1.阅读论文《BAE: BERT-based Adversarial Examples for Text Classification》

2.《Adversarial Text Generation by Search and Learning》

3.《TextHacker: Learning based Hybrid Local Search Algorithm forText Hard-label Adversarial Attack》

4.《Query-Efficient and Scalable Black-Box Adversarial Attacks onDiscrete Sequential Data via Bayesian Optimization》

5.《Experience speaks louder: Black-box hard-label adversarial attackthrough reinforcement learning》

本周

问题汇报

1.暂无

下周工作计划

1.对已有的idea进行验证和实验

2. 继续阅读更多毕业设计相关论文