Events

Filters
Export this event to calendar
iCal
RSS
RSS
Close X

Filter by:

Limit to events where the title matches:
Limit to events where the first date of the event:
Date range
Limit to events where the type is one or more of:
Limit to events tagged with one or more of:
Limit to events where the audience is one or more of:
Tuesday, September 23, 2025 10:30 am - 11:30 am EDT (GMT -04:00)

Joint Distinguished Lecture by Bin Yu

Dean's Distinguished Women in Mathematics, Statistics and Computer Science Lecture Series & David Sprott Distinguished Lecture Series

Bin Yu
CDSS Chancellor's Distinguished Professor, Statistics, EECS, Center for Computational Biology
Senior Advisor, Simons Inst for the Theory of Computing
Member, U.S. National Academy of Sciences, 2014
Member, American Academy of Arts and Sciences, 2013
Guggenheim Fellow, 2006

Room: DC 1302

Veridical Data Science towards Trustworthy AI 

In this talk, I will introduce the Predictability-Computability-Stability (PCS) framework for veridical (truthful) data science, highlighting its critical role in producing reliable and actionable insights. I will share success stories from cancer detection and cardiology, showcasing how PCS principles have guided cost effective designs and improved outcomes in these projects. Since trustworthy uncertainty quantification is indispensable for trustworthy AI, I will discuss PCS uncertainty quantification for prediction in regression and multi-class classification. PCS-UQ consists of three steps: pred-check, bootstrap, and multiplicative calibration. Through test results over 26 benchmark datasets, PCS-UQ will be shown to outperform common forms of conformal prediction in terms of width, subgroup coverage, and subgroup interval width. Finally, the multiplicative step in PCS-UQ will be shown to be a new form of conformal prediction.