Building ML for All
How do we build ML that works well for everyone? In recent years, ML has increasingly shifted from precise, well-defined tasks to diverse use cases. This presents new challenges for how to understand where these systems work well or don't, and how to improve them. In this talk, I will dive deep into how we've made progress in addressing these challenges in recommender systems. First, I'll discuss progress in measuring and improving ML Fairness in recommender systems, across diverse users and item providers. Second, I'll discuss research on leveraging reinforcement learning for all users to have positive, long-term experiences. Last, building on our understanding from recommenders, I'll discuss how such challenges of robust performance arise in language models, and the challenges and opportunities in this area.
Alex Beutel is a Senior Staff Research Scientist and team lead in Google Research, driving research spanning recommender systems, fairness, robustness, safety, reinforcement learning, and ML for databases. He received his Ph.D. in 2016 from Carnegie Mellon University’s Computer Science Department, and previously received his B.S. from Duke University in computer science and physics. His Ph.D. thesis on large-scale user behavior modeling, covering recommender systems, fraud detection, and scalable machine learning, was given the SIGKDD 2017 Doctoral Dissertation Award Runner-Up. He received the Best Paper Award at KDD 2016 and ACM GIS 2010, was a finalist for best paper in KDD 2014, and was awarded the Facebook Fellowship in 2013 and the NSF Graduate Research Fellowship in 2011. More details can be found at alexbeutel.com.