SACRO-ML
SACRO-ML copied to clipboard
AI-SDC
Collection of tools and resources for managing the statistical disclosure control of trained machine learning models
At moment we effectively run a worst-case attack where a simulated attacker has the model which outputs probabilities, and has a record with the target label and with just the...
Sub-class of safeModelClassifier: - needs parameter recommendations. from @rpreen 's results - ideally should implement the k-anonymity reporting functionality - but recognise I don't know if their forest/tree structure is...
probably need to create a list of what attributes are expected to be present then simply check keys in self.__dict__ against that list
updates: - [github.com/astral-sh/ruff-pre-commit: v0.5.7 → v0.6.1](https://github.com/astral-sh/ruff-pre-commit/compare/v0.5.7...v0.6.1)
![pre-commit-ci[bot] avatar](https://avatars.githubusercontent.com/in/68672?v=4 "pre-commit-ci[bot] avatar"){kind=avatar}
Should the preprocessing data loaders be part of the `aisdc` package? It doesn't seem likely that many people will want to use them and they are more suited to going...
As of `1.2.0` the CLI prompt tools were added: * `$ aisdc gen-attack` tool that produces the attack config `attack.yaml` for running attacks on the CLI * `$ aisdc gen-target`...
then sanity it check it with various TREs - maybe even to sdc-reboot list