Course Objective & Scope
When banks, hospitals, group subsidiaries or edge devices want to train a shared fraud, risk or forecasting model, data protection laws and commercial confidentiality often forbid pooling raw records centrally. This course covers cross-silo and cross-device federated learning architectures, orchestration with the Flower framework, FedAvg, FedProx and FedAdam algorithms, and techniques for handling heterogeneous (non-IID) client data.
To prevent leakage from model updates (gradient inversion, membership inference), Secure Aggregation and differential privacy with Opacus (DP-SGD) are practised hands-on. Participants build a multi-silo simulation prototype and report the trade-off between accuracy, communication cost and privacy budget.
Who Should Attend?
- Data scientists, ML engineers and platform engineers
- Teams building privacy-preserving AI in finance, healthcare, telecom and enterprise groups
- AI security and data governance architects
- Prerequisite: Python and model training experience with PyTorch or scikit-learn
Key Takeaways
- Select horizontal or vertical federated architectures for cross-silo and cross-device use cases
- Build multi-client FedAvg and FedProx training loops with the Flower framework
- Apply local personalisation and fine-tuning strategies under non-IID client distributions
- Protect model updates with Secure Aggregation and differential privacy (Opacus DP-SGD)
- Benchmark centralised, local-only and federated models across accuracy, bandwidth and epsilon budgets
Curriculum
Federated Architectures and Enterprise Use Cases
- Cross-silo vs cross-device; horizontal, vertical and federated transfer learning
- Data residency, GDPR/KVKK constraints and multi-party collaboration models
Federated Training and Aggregation with Flower
- Client-server loops, FedAvg, FedProx and FedOpt strategies
- Federated setups for PyTorch networks and tabular/GBDT models
Heterogeneous (Non-IID) Data and Personalisation
- Distribution skew across silos and mitigating client drift
- Local fine-tuning, split heads and clustered federated learning
Privacy Attacks and Differential Privacy
- Gradient inversion, membership inference and poisoning threats
- DP-SGD with Opacus, clipping, epsilon-delta accounting and Secure Aggregation
Operations, Communication Cost and Governance
- Update compression, straggler handling and asynchronous rounds
- Contribution valuation, audit trails and the federated evaluation report
Other Courses in This Category
-
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
-
Supervised and unsupervised learning, train/test splits, overfitting, core metrics and first models in scikit-learn, so technical teams can frame a business problem as an ML task and build a sound baseline.
- Live Course
- 5 modules
-
Gradient boosting, ensembles and stacking, hyperparameter optimisation with Optuna, imbalanced data, probability calibration and custom loss functions for experienced data scientists building production-grade models.
- Live Course
- 5 modules
Get in touch
