Designing Recommender Systems: A Case Study Approach from Discriminative to Generative AI
In today’s digital economy, AI-driven recommendation systems (RecSys) are at the heart of successful businesses, powering personalization in e-commerce, media, fintech, and beyond. From boosting customer engagement to increasing revenue through targeted recommendations, organizations rely on machine learning-powered RecSys to drive business outcomes. This course provides a practical, industry-focused approach to designing and deploying next-generation recommender systems. Participants will explore the evolution from traditional (discriminative) to cutting-edge (generative) AI approaches, gaining hands-on experience with collaborative filtering, deep learning-based recommendations, and generative AI for hyper-personalization.
Through real-world case studies in e-commerce, streaming services, and healthcare, attendees will learn how to:
- Build scalable and business-driven recommender systems
- Optimize personalization strategies to increase engagement and revenue
- Address bias, fairness, and ethical concerns in AI-powered personalization
- Leverage Generative AI for dynamic content and user experience enhancement
Content
Introduction to Personalization and Recommender Systems
- Collaborative Filtering,
- Content-based Filtering,
- Hybrid
Computational methods for designing Recommender Systems (Discriminative modelling)
- Learning representations from data
- Unimodal and Multimodal representation learning
- Transfer Learning
- Neural Topic Modelling
- Matrix Factorization and Singular Value Decomposition (SVD)
Introduction to Generative AI for Recommender Systems:
- Autoencoders
- Variational Autoencoders (VAEs),
- Generative Adversarial Networks( GANs)
Introduction to Reinforcement Learning (RL) for Recommender Systems
- Contextual Bandits for Personalized Recommendations
- Multi-Armed Bandit Problem and Real-Time Recommendations
Human-Centered Recommendation Systems
- Why Human-centered RecSys
- Human-centred RecSys design pipeline
- Multi-stakeholder awareness in Recommender systems
- Context-aware recommender systems
Evaluating Recommender systems
- Evaluation Metrics
- Offline Evaluation
- Conducting a User study
- Online Evaluation
Learning Outcomes
- Develop foundational knowledge of Recommender Systems (RecSys), including both discriminative and generative approaches.
- Understand a wide variety of RecSys algorithms and their application across diverse domains, including healthcare and visual arts.
- Design and evaluate human-centred RecSys, incorporating principles of user experience, ethical AI, and interdisciplinary considerations.
- Apply practical skills to design, implement, and evaluate RecSys using real-world datasets, with a focus on personalization and innovative solutions.
- Analyze case studies to understand the role of RecSys in addressing challenges and propose solutions
Training Method
An intensive week with theory in the mornings and hands-on in the afternoons. This balanced approach helps participants understand both the technical details and business impact, giving them practical and useful insights to apply AI-powered personalization in their careers or workplaces.
Organised By
Digital Learning Hub Luxembourg
Certification
Participation OnlyPrerequisites
- Familiarity with Machine Learning
- Knowledge of Algebra and Calculus
- Prior experience with Python programming language
Planning and location
09:00 - 17:00
09:00 - 17:00
09:00 - 17:00
09:00 - 17:00
09:00 - 17:00
Your trainer(s) for this course
