DataScience@BI seminar with Geir Kjetil Sandve
DataScience@BI seminar invites Professor of Scientific Computing and Machine Learning Geir Kjetil Sandve to give a talk about "Open and truthful science".
- Starts:12:00, 13 October 2026
- Ends:13:00, 13 October 2026
- Location:BI - campus Oslo, B3 inner area next to meeting room B3i-108
- Enrolment deadline:13.10.2026 11:00
- Contact:Siri Johnsen (siri.johnsen@bi.no)
Abstract:
Reproducibility has become an established cornerstone of trustworthy data science: sharing code, data and computational environments lets others rerun an analysis and obtain the same results. However, reproducibility is necessary but not sufficient for research to be trustworthy; most findings rest on multiple judgment calls that could potentially have led to other conclusions if made differently. This ranges from judgement on how data is cleaned to how a statistical models are chosen and how performance is assessed. The recently proposed Veridical Data Science framework advocates that transparency should therefore extend beyond code and data to the judgment calls themselves, and to how stable the conclusions are across reasonable alternatives. This is particularly pressing in the current era where Agentic AI is disrupting science, which may multiply the current challenges with fragile findings, but may also open for new solutions. I will in the talk introduce the motivation for and approaches to open and truthful science, and illustrates this through cases from my own research on deep learning methodology to decipher the immune system and on spatio-temporal modelling for early warning of disease outbreaks.
The DataScience@BI seminar is organised by the Department of Data Science and Analytics, BI Norwegian Business School.
Visit our website to read more about the Department of Data Science and Analytics.