Machine learning for collaborative intelligent systems

The course aims to provide students with an understanding of machine learning algorithms and architectures required to realize adaptive collaborative intelligent systems and to familiarize them with the methods to evaluate such systems. Students will learn how to create systems that produce effective collaborative behaviors and are robust to human presence and their complex behaviors. These systems should implement "empathetic functions" to estimate, reason, and learn about human-specific quantities such as intentions, beliefs, preferences, emotional and attentional states, individual differences, and cognitive bounds and variations in performance. Examples of such systems include autonomous driving cars, voice assistants, and recommender systems, utilizing algorithms like reinforcement learning, federated learning, and inverse reinforcement learning.

After successfully completing the course, students will be able to:
· Understand and apply machine learning algorithms and architectures for the creation of collaborative intelligent systems.
· Describe and implement methods to make systems robust to human presence and capable of producing effective collaborative behaviors.
· Select, implement, and apply "empathetic functions" that will allow collaborative systems to estimate, reason, and learn about human-specific quantities such as intentions, beliefs, preferences, emotional and attentional states, individual differences, and cognitive performance variations.
· Define and implement communication and guidance within collaborative systems.
· Incorporate human partners' satisfaction and wellbeing into the objective functions of intelligent systems.
· Apply specific algorithms, such as reinforcement learning for behavior adaptation, federated learning for distributed model training, and inverse reinforcement learning for understanding human actions and goals.
· Evaluate adaptive collaborative intelligent systems based on objectives such as usability, visibility, interpretability, effectiveness, goals, security, and privacy.
This comprehensive understanding will enable students to design, evaluate, and implement AI systems that effectively interact with and support human users, as seen in applications like autonomous vehicles, smart assistants, and personalized recommendation engines.