Model risks Architecture Patterns

RCCE students will learn machine learning model security risks including adversarial attacks, model poisoning, model theft, model inversion, and membership inference attacks. RCCE students will learn to assess ML model security throughout the model lifecycle from training through deployment, identify vulnerabilities in model architectures and training pipelines, detect adversarial input attacks designed to cause misclassification, prevent model poisoning through training data integrity controls, protect model intellectual property against extraction attacks, implement model monitoring for drift and adversarial behavior, and develop incident response procedures for compromised ML models. This architecture course teaches secure system design using proven patterns, guardrails, and reference architectures. At an expert level, RCCE students will learn to evaluate design options against security requirements, make informed trade-off decisions, and build systems that are resilient by design. Students gain the architectural thinking skills needed for security engineering and solution design roles.