Course Objective & Scope
Beyond basic SHAP plots, high-stakes credit, insurance, HR or healthcare models demand mathematical fidelity, robustness under correlated features and actionable recourse for affected users. This advanced programme covers inherently interpretable glass-box models (InterpretML Explainable Boosting Machine, GAMs), interventional vs conditional SHAP, Accumulated Local Effects (ALE) in place of PDP when features correlate, and feasible counterfactual explanations with DiCE.
It also addresses Integrated Gradients for neural models, conformal prediction for calibrated uncertainty sets and right-to-explanation obligations under the EU AI Act and data protection rules. Participants design an audit-ready XAI package and multi-audience explanation templates for a high-risk model.
Who Should Attend?
- Senior data scientists and ML engineers
- Model validation, model risk and AI governance specialists
- Teams building finance, insurance, healthcare and public-sector decision systems
- Prerequisite: ML modelling experience and familiarity with basic evaluation/SHAP
Key Takeaways
- Build glass-box models such as InterpretML EBMs and constrained GBDTs without sacrificing accuracy
- Diagnose SHAP and PDP pitfalls under correlated features and apply ALE and grouped explanations
- Generate feasible, constraint-aware counterfactual recourse with DiCE
- Attach statistically valid prediction sets to individual decisions using conformal prediction
- Standardise layered explanation outputs for end users, operations teams and auditors
Curriculum
Inherently Interpretable Glass-Box Models
- Generalised Additive Models (GAMs) and InterpretML EBMs
- Explainability by design with monotonic and interaction-constrained models
Shapley Theory and the Correlation Problem
- TreeSHAP, KernelSHAP, interventional vs conditional expectations
- PDP failure modes under correlated features and ALE alternatives
Counterfactual Explanations and Algorithmic Recourse
- Generating diverse, plausible counterfactuals with DiCE
- Immutable attributes, actionability and causal feasibility constraints
Deep Models and Uncertainty Explanations
- Integrated Gradients, attention and prototype-based explanations
- Conformal prediction with MAPIE and abstention policies
Regulation, Audit and Explanation Robustness
- Testing explanation fidelity, stability and adversarial manipulation
- EU AI Act high-risk documentation and adverse-action reason codes
Other Courses in This Category
-
Moving from correlation to causation with DAGs, propensity scores, Double Machine Learning (EconML/DoWhy), heterogeneous treatment effects (CATE) and Difference-in-Differences to measure price, campaign and policy impact.
- Live Course
- 5 modules
-
Collaborative model training across institutions or edge devices without centralising raw data: Flower framework, FedAvg/FedProx, non-IID data, differential privacy (Opacus) and secure aggregation.
- Live Course
- 5 modules
-
Intro to Machine Learning - Regression Models
Hands-on workshop covering Linear, Ridge, Lasso, and Polynomial regression models, feature engineering, MSE/RMSE/MAE/R² evaluation metrics, and regularization techniques.
- Live Course
- 3 modules
-
Incremental learning on streaming data with River, prequential evaluation, concept drift detection (ADWIN, Page-Hinkley) and contextual bandits for continuous adaptation in real-time systems.
- Live Course
- 5 modules
-
Modelling relational data with NetworkX and PyTorch Geometric: centrality and community algorithms, graph features, Node2Vec and GCN/GraphSAGE/GAT for node, link and subgraph prediction in fraud rings and supply networks.
- Live Course
- 5 modules
-
In-context learning on tabular data with TabPFN v2, plus zero-shot and fine-tuned time-series forecasting with Chronos, TimesFM, Moirai and Lag-Llama, benchmarked against GBDTs and classical models.
- Live Course
- 5 modules
Get in touch
