Differentially Private Continual Learning
Differentially Private Continual Learning
复制标题
差异化私人持续学习
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Y. Gal
中科院分区：
文献类型：
--
作者：
Sebastian Farquhar;Y. Gal
对于出于隐私原因必须删除历史数据的机构来说，灾难性的遗忘可能是一个严重的问题。例如，医院可能无法永久保留患者数据。但是，仅根据最近的数据训练的神经网络往往会忘记从旧数据中吸取的教训。提出了一种基于变分推理的差分私有持续学习框架。在给定当前模型的情况下，我们使用旧数据集的不同私有生成模型来估计过去数据的可能性。
Catastrophic forgetting can be a significant problem for institutions that must delete historic data for privacy reasons. For example, hospitals might not be able to retain patient data permanently. But neural networks trained on recent data alone will tend to forget lessons learned on old data. We present a differentially private continual learning framework based on variational inference. We estimate the likelihood of past data given the current model using differentially private generative models of old datasets.