Saltar al contenido principal

Curso completo próximamente

Distributed Model Training: Privacy, Architecture, and Coordination

7 capítulos · 31 lecciones

  1. 1. Distributed Training Architectures and When to Use Them

    4 lecciones

    Compare centralized, federated, and hybrid training patterns and map architectural choices to privacy, latency, and compliance requirements.

  2. 2. Federated Learning Protocols and Coordination

    5 lecciones

    Implement the core federated averaging algorithm and build client selection strategies that handle device heterogeneity and availability.

  3. 3. Non-IID Data and Statistical Heterogeneity

    4 lecciones

    Address the core challenge of federated learning: training effective models when each client has fundamentally different data distributions.

  4. 4. Privacy Mechanisms: Secure Aggregation and Differential Privacy

    5 lecciones

    Implement cryptographic and statistical privacy techniques that protect individual client contributions during model aggregation.

  5. 5. Security, Robustness, and Malicious Client Detection

    4 lecciones

    Build defenses against poisoning attacks, Byzantine clients, and model inversion attempts in open federated systems.

  6. 6. Deployment Patterns for Mobile, Edge, and Enterprise

    5 lecciones

    Deploy federated learning systems across mobile devices, IoT infrastructure, and enterprise data silos with real-world constraints.

  7. 7. Monitoring, Debugging, and Production Operations

    4 lecciones

    Build observability systems and operational practices for federated learning pipelines that span thousands of distributed clients.

Recibe un correo cuando abra este curso