Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn the fundamental principles of interactive decision making in machine learning through this 58-minute conference talk that presents a comprehensive framework for understanding how learning agents operate in dynamic environments. Explore the critical exploration-exploitation dilemma that arises when machine learning methods are deployed in interactive settings, from dynamic treatment strategies in healthcare to reinforcement learning applications in large language model fine-tuning. Examine a unified theoretical framework that encompasses multi-armed bandits, contextual bandits, structured bandits, and reinforcement learning as special cases. Delve into the statistical foundations of learning in interactive environments, focusing on developing precise characterizations of sample complexity based on model class properties. Discover the essential algorithmic primitives that underpin effective interactive decision making systems and understand how these components work together to enable intelligent exploration and exploitation strategies in real-world applications.
