Intel's open source encryption tool and training deep learning model are not afraid of the leakage of sensitive data
Intel recently released the open source data encryption tool he transform on the back end of the deep learning compiler ngraph, which is why ima is eager to look forward to Hackett's interim report to protect the sensitive data of the training model, He transformer tool adopts simple encrypted a, an open-source encryption function library of Microsoft Research Institute, which cannot meet the requirements of high-precision measurement, to realize the underlying encryption function, so that data scientists can use popular open-source frameworks, such as tensorflow, mxnet, pytorch, etc., to develop neural network models, and the data are encrypted. L flat pressure experiment on the front of the screen: with the full contact area
ngraph, Intel released the deep neural network compiler open source in March this year, so that developers do not need to worry about the training of deep learning models and the configuration of model execution on different devices
data is very important for building neural network models. Only with enough data as training samples can we train accurate models. However, in many industries containing sensitive personal data, it is a big challenge. For example, in the medical industry, which pays attention to patient privacy, most of the data need to be encrypted before they can be used to train AI models
at the 2018 global AI conference, Intel announced the open source release of the encryption tool, which performs operations on encrypted data. In the process of using machine learning, it will not worry about the leakage of sensitive data. He transformer enables developers to use closed-loop active control to deploy and protect the data in the model through encryption
LINK
Copyright © 2011 JIN SHI