Explainable AI for Industry

• Chapter 1.1: What is Explainable AI?

• Chapter 1.2: Why Explainability Matters in Industry

• Chapter 1.3: Regulatory and Ethical Drivers (GDPR, AI Act)

• Chapter 2.1: Model-Specific vs. Model-Agnostic Techniques

• Chapter 2.2: Global vs. Local Explanations

• Chapter 2.3: Common Techniques: SHAP, LIME, Anchors, Partial Dependence

• Chapter 3.1: Integrating SHAP in Tree-based Models

• Chapter 3.2: Using LIME for Text and Image Classification

• Chapter 3.3: Visualizing Feature Importance and Decision Boundaries

• Chapter 3.4: Performance vs. Interpretability Trade-Offs

• Chapter 4.1: Overview of XAI Libraries (SHAP, LIME, ELI5, Captum)

• Chapter 4.2: Explainability in Deep Learning with Captum

• Chapter 4.3: Explainability in Production with MLflow and dashboards

• Chapter 4.4: Case: Interpreting Drift and Model Updates in Production

• Chapter 5.1: Financial Services – Credit Risk, Fraud Detection

• Chapter 5.2: Healthcare – Diagnostics, Clinical Decision Support

• Chapter 5.3: Manufacturing – Predictive Maintenance, QA Automation

• Chapter 6.1: Model Auditing and Documentation Practices

• Chapter 6.2: Explaining AI to Non-Technical Stakeholders

• Chapter 6.3: Building Trustworthy AI Pipelines

• Chapter 6.4: Capstone – Develop an XAI Dashboard for Business Decision Makers

• Introduction to Explainable AI (15 min)

• History and Motivation for XAI (30 min)

• Global vs. Local Explanations (30 min)

• Model-Specific vs. Model-Agnostic Methods (30 min)

• SHAP and LIME Basics (30 min)

• Anchors, PDP, ICE Plots Overview (30 min)

• Ethics, Bias, and Compliance in AI (45 min)

• GDPR and AI Act: Regulatory Impacts on ML Pipelines (45 min)

• Visualizing Feature Importance (30 min)

• SHAP in Tree Models (Hands-On) (45 min)

• LIME for Text Classification (Hands-On) (45 min)

• XAI for Image Classification with Grad-CAM & LIME (45 min)

• Using Captum for Deep Learning Models (30 min)

• Trade-Offs: Performance vs. Interpretability (30 min)

• Production Tools: MLflow, ELI5, Explainability Dashboards (45 min)

• Monitoring for Model Drift and Explanation Validity (45 min)

• Interpretable Pipelines for Regulated Environments (30 min)

• Healthcare XAI: Clinical Use Case Deep Dive (45 min)

• Manufacturing XAI: QA and Maintenance Models (30 min)

• Communicating AI Decisions to Executives (30 min)

• Creating an XAI Report for Audit and Stakeholders (45 min)

• Capstone Guide: Building an XAI Dashboard (60 min)

• Capstone Showcase Examples (30 min)

• Final Review and Next Steps in Responsible AI (30 min)

• Future Trends: Glass-Box Models and Causal Explainability (15 min)

  ---

Title: Beyond the Black Box: The Why and How of Explainable AI
Duration: 60 minutes
Focus: Fundamentals of XAI, business drivers, and regulatory expectations
Guest: AI Ethicist / Responsible AI Lead
Interactive: Live group activity using SHAP and LIME visualizations on a simple ML model

Title: Interpretable Models in Practice: From Prediction to Justification
Duration: 75 minutes
Focus: Practical integration of explainability tools into ML workflows
Guest: Machine Learning Engineer / XAI Researcher
Interactive: Model debugging session with SHAP, LIME, and Captum applied to tabular and image data

Title: Explainability in Production: Communicating AI to Stakeholders
Duration: 90 minutes
Focus: Real-world deployment, stakeholder communication, and compliance documentation
Guest Panel: Data Scientist + UX Designer + Policy Analyst
Interactive: Case-based presentation critique + live Q&A on explainability reports and dashboards

Edit

Edit