Fractal: Empowering better decisions Remarkably, every typical AI life cycle stage presents opportunities and requirements for incorporating RAI modules. There is no stage where RAI modules can be disregarded, and it's crucial to embed RAI into the system right from the start rather than adding it as an afterthought once the AI development process has started.
Data science lifecycle with RAI
EDA, Pre-processing and feature engineering and selection
Business understanding and hypothesis testing
Model development
Model evaluation
Model selection
Input data
RAI definition
Data privacy
Data bias
XAI & privacy
Model bias & privacy
Model management
Model accountability
Deployment
Prediction
Monitoring
Data privacy
XAI, Prediction bias
Bias, XAI & Drifts
Applying our enablers, we have developed a comprehensive and reusable toolkit to integrate RAI into clients' data science life cycles seamlessly. Our codified frameworks, policies, guidelines, and cloud-agnostic codes can be applied across any data and data science process. By democratizing these resources through educational programs, we empower clients to effectively embed RAI principles and practices into their workflows and organizational practices.
Our toolkit includes the following components:
Integrating the RAI framework and its underlying principles into AI systems to ensure responsible practices are integrated from the start. 1. EMBEDDING RAI FRAMEWORK Providing comprehensive guidelines and recommendations to guide clients in implementing RAI effectively. 2. GUIDELINES AND RECOMMENDATIONS:
© 2023 Fractal Analytics Inc. All rights reserved
10
Made with FlippingBook - PDF hosting