DeepMemory: Model-based Memorization Analysis of Deep Neural Language Models

2021 36th IEEE/ACM International Conference on Automated Software Engineering (ASE)(2021)

引用 4|浏览14
暂无评分
摘要
The neural network model is having a significant impact on many real-world applications. Unfortunately, the increasing popularity and complexity of these models also amplifies their security and privacy challenges, with privacy leakage from training data being one of the most prominent issues. In this context, prior studies proposed to analyze the abstraction behavior of neural network models, e.g...
更多
查看译文
关键词
Analytical models,Data privacy,Recurrent neural networks,Semantics,Training data,Markov processes,Data models
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要