IS671 — Responsible AI¶

Building fair, explainable, accountable, and privacy-preserving AI systems.

This wiki covers the core concepts, techniques, and Python implementations from Responsible AI: Implementing Ethical and Unbiased Algorithms by Sray Agarwal & Shashin Mishra (Springer, 2021) — as taught at California State University, Long Beach.

The Four Pillars of Responsible AI¶

graph LR
    RAI[Responsible AI] --> F[⚖️ Fairness]
    RAI --> X[🔍 Explainability]
    RAI --> A[📊 Accountability]
    RAI --> P[🔒 Privacy]
    F --> F1[Bias Detection]
    F --> F2[Bias Mitigation]
    X --> X1[Feature Explanation]
    X --> X2[Model Explanation]
    A --> A1[Data Drift]
    A --> A2[Concept Drift]
    P --> P1[Differential Privacy]
    P --> P2[Federated Learning]

Course Chapters¶

Ch	Topic	Focus
1	Introduction	What is RAI and why it matters
2	Fairness & Proxy Features	Confusion matrices, fairness metrics, proxy detection
3	Bias in Data	Statistical parity, disparate impact, visualization
4	Explainability	Feature, model & output explanation techniques
5	Remove Bias from ML Model	Reweighting, counterfactual fairness
6	Remove Bias from Output	Reject Option Classifier
7	Accountability	Data drift, concept drift, monitoring
8	Data & Model Privacy	Differential privacy, federated learning, attacks
9	Conclusion	RAI lifecycle, canvas, and checklist

How to Navigate¶

Each chapter builds on the previous:

Understand fairness — what it means, how to measure it (Ch 1–2)
Detect bias in your data before training (Ch 3)
Explain why your model makes decisions (Ch 4)
Fix bias at the model level and output level (Ch 5–6)
Monitor for drift and degradation in production (Ch 7)
Protect data and model privacy (Ch 8)

Running the Code

All Python examples use standard data science libraries: scikit-learn, pandas, numpy, matplotlib, and seaborn. Install them with:

pip install scikit-learn pandas numpy matplotlib seaborn

Course: IS671 — Responsible AI | CSULB | Prof. Jose Pineda Textbook: Agarwal & Mishra, Responsible AI (Springer, 2021)