MACHINE LEARNING (2/1 Credits)

Learning Outcomes:

On successful completion of this course, students will be able to: LO1 – differentiate between traditional deterministic software and probabilistic machine learning systems, identifying the unique challenges in requirements, testing, and deployment; LO2 – assess data quality by detecting noise, handling missing values, and applying appropriate feature engineering techniques for numerical, text, and image data; LO3 – design automated machine learning pipelines (MLOps) that integrate data collection, model training, and validation into a cohesive software architecture; LO4 – test machine learning components using unit and integration testing frameworks to manage the non-deterministic nature of AI models; LO5 – Deploy machine learning models as scalable web services and microservices using tools like Flask and Docker within a larger software ecosystem; LO6 – evaluate ethical risks, algorithmic bias, and legal obligations in data acquisition to ensure the development of responsible and fair software systems.

Topics:

1. Data Quality and Feature Engineering;
2. Designing Automated ML Pipelines;
3. Engineering with Pre-Trained Models (Hugging Face);
4. Testing Machine Learning Systems;
5. Model Deployment with Flask and Docker;
6. Ethics and Bias Detection;
7. Machine Learning vs. Traditional Software Engineering;
8. Architecture of Production ML Systems;
9. Data Acquisition and Types in Software Systems;
10. Data Quality and Noise Management;
11. Feature Engineering: Numerical and Image Data;
12. Feature Engineering: NLP and Source Code;
13. Designing ML Models: Feature-Based vs. Deep Learning;
14. Training Advanced Models (GPT and Autoencoders);
15. Designing MLOps Pipelines;
16. Testing and Monitoring ML Systems;
17. Deployment and Ecosystem Integration;
18. Ethics, Bias, and Legal Obligations;
19. Project Presentation.