Bernoulli Lecture - Veridical Data Science   24 June 2020

Part of the Semester : Functional Data Analysis

17:15 - 18:15
Room : BCH 2201

Lecturer

Bin Yu, University of California, Berkeley

Veridical data science extracts reliable and reproducible information from data, with an enriched technical language to communicate and evaluate empirical evidence in the context of human decisions and domain knowledge. Building and expanding on principles of statistics, machine learning, and the sciences, we propose the predictability, computability, and stability (PCS) framework for veridical data science. Our framework is comprised of both a workflow and documentation and aims to provide responsible, reliable, reproducible, and transparent results across the entire data science life cycle. Moreover, we propose the PDR desiderata for interpretable machine learning as part of veridical data science (with PDR standing for predictive accuracy, predictive accuracy and relevancy to a human audience and a particular domain problem).

The PCS framework will be illustrated through the development of iterative random forests (iRF) for extracting preditable and stable non-linear interactions in genomics studies. Finally, a general DNN interpretaion method based on contexual decomposition (CD) will be discussed with applications to sentiment analysis and cosmological parameter estimation.

Name University Dates of visit
Bin Yu University of California, Berkeley 24/06/2020 - 24/06/2020
Total Guests : 1
Name University Dates of visit
Total Guests : 0
Conference in Honor of the 70th Birthday of Tudor Ratiu, 20 to 24 July 2020.
Registration fees of CHF 50.- are mandatory for visits from 1 day to full attendance of the conference.
Cancellations and no-shows are not eligible for a refund.
I hereby agree to the conditions of the registration fees and to settling the amount by 28 February 2020.

Last Name
First Name
Address
Zip Code
City
Country
Email
Participation
CHF