Optimizing Recommendations with Multi-Armed & Contextual Bandits for Personalized Next Best Actions

In this WiDS Upskill Workshop, Keerthi Gopalakrishnan explores how Multi-Armed Bandit (MAB) and Contextual Bandit algorithms can optimize online recommendations for next-best-action scenarios. These techniques help balance the trade-off between exploration (trying new recommendations) and exploitation (leveraging successful actions) to drive better personalization and engagement. Keerthi will break down key MAB concepts, including: Epsilon-greedy Upper Confidence Bound (UCB) & Contextual UCB Thompson Sampling Real-world applications in recommendation systems This session is ideal for: Data Scientists and Machine Learning Engineers Product Managers and Data Strategists Researchers and Academics Prior knowledge: A background in supervised learning and evaluation metrics is recommended. Familiarity with online learning or decision-making algorithms is helpful but not required. Learn more about WiDS Upskill Workshops: https://www.widsworldwide.org/learn/u...