Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Learn how to address predictive multiplicity challenges in machine learning systems through this 13-minute conference talk from ACM's Fairness Metrics and Technical Approaches session. Explore practical approaches for reconciling situations where multiple equally valid models produce different predictions for the same input, examining the implications for fairness and system deployment. Discover evaluation practices and methodologies that researchers from Stony Brook University, University of Oxford, CMU Tepper School of Business, and Emory University have developed to handle predictive multiplicity in real-world machine learning applications. Gain insights into system development strategies that account for model uncertainty and learn how to implement robust evaluation frameworks when dealing with multiple viable predictive models.
