Support Centre

You have out of 5 free articles left for the month

Signup for a trial to access unlimited content.

Start Trial

Continue reading on DataGuidance with:

Free Member

Limited Articles

Create an account to continue accessing select articles, resources, and guidance notes.

Free Trial

Unlimited Access

Start your free trial to access unlimited articles, resources, guidance notes, and workspaces.

USA: NIST publishes blog on machine learning with differential privacy

The National Institute of Standards and Technology ('NIST') published, on 21 December 2021, a blog as part of its differential privacy blog series on machine learning. In particular, NIST noted that machine learning is increasingly being used for sensitive tasks, such as medical diagnosis. In addition, NIST outlined that differentially private machine learning algorithms can be used to quantify and limit leakage of private information from a learner's training data. Moreover, NIST highlighted that machine learning allows trainers to prevent the memorisation of their training set, such as the specific medical histories of individual patients, as well as allowing them to responsibly train models on sensitive data.

Moreover, the blog covers the following areas:

  • why we need private machine learning algorithms;
  • differential privacy helping to design better machine learning algorithms;
  • private algorithms for training deep learning models;
  • differentially private stochastic gradient descent;
  • model agnostic private learning; and
  • practical deployment and software tools.

You can read the blog here.