推荐singhrac · 2018-03
“and builds on this great paper: [2].”
称它是这项工作所基于的优秀论文
在差分隐私框架下训练深度神经网络,并精细核算隐私代价
针对训练数据可能含敏感信息、模型不应泄露隐私的问题,提出新的学习算法技术,并在差分隐私框架下对隐私代价做更精细的分析。实验表明,可以在适度的隐私预算下训练带非凸目标的深度神经网络,在软件复杂度、训练效率和模型质量上的代价都可接受。
评论者把它当作学习“如何在差分隐私保证下训练神经网络”的入门起点,并称后续工作建立在这篇论文之上。
“and builds on this great paper: [2].”
“A good place to start on the topic is Abadi 2016.”
“My former coworker Li Zhang has an Erdos number of 2 and we worked together on this paper: Deep Learning with Differential Privacy (with him doing a lot more work than me).”