Skip to content

schelterlabs/projects-amnesia

Repository files navigation

Amnesia

To comply with laws, which protect the “right to be forgotten” and delete customer data from ML models (considered to be a lossy compressed version of customer data), Amnesia enables users to decrementally update trained ML models (for recommendation, classification, and regression), thereby forgetting the data of a customer without necessitating model retraining.

See our publication for details: https://ssc.io/pdf/amnesia-cidr.pdf

About

Code and experiments for our CIDR paper on "A Selection of Machine Learning Models that Can Forget User Data Very Fast"

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published