All 3rd Party tools for Responsible AI

Responsible AI; RAI is umbrella term for a trustworthy AI models. It includes-

1. Fairness of Data and model ( gender, racial biases)

2. Explainability of Data and model ( local and global interpretability and explainability of models)

3. Robustness ( Security and safety of model and data)

4. Soundness ( Data Quality and model performance)

5. Accountability ( Governance and compliance)

6. Privacy ( Data Ethics and human consent)

7. Sustainability( Societal and Environment well- being)

There are multiple frameworks available to test above mentioned components of RAI. Here we list few-

VENDOR	framework/package name	RAI Component
IBM	Watson Openscale	Fairness and Transparency :Analyze the asset/solution with trust and transparency and understand how the model makes decision. Automate AI at scale with transparent, explainable outcomes that are free from harmful bias and drift
​IBM	AI Fairness 360	Fairness: Examine, report, and mitigate discrimination and bias in ML models
IBM	AI Explainability 361	Explainability and Interpretability
IBM	Diffprivlib	Privacy: explore the impact of differential privacy on ML accuracy using classification and clustering models
Google	What-If Tool	Explainability and Interpretability: using visualization your dataset automatically
Google	Explainable AI Tool	Explainability : Design and build interpretable AI
Google	Differential Privacy Library	Privacy
Microsoft	InterpretML	Transparency
Microsoft	Fairlearn	Fairness
Microsoft & OpenDP	SmartNoise	Privacy
Amazon	Sagemaker Clarify	Transparency
DataRobot	Bias & Fairness Testing	Fairness
H20.ai	H20 Driverless AI	Transparency
Diveplane	Understandable AI	Transparency
Diveplane	GEMINAI	Privacy
Facebook & Pytorch	Opacus	Privacy
DataRobot	Explainable AI	Transparency
LinkedIn	LiFT	Fairness
Fairkit-Learn	Fairkit-Learn	Fairness
Aequitas	Bias & Fairness Audit	Fairness