Our fifth workshop is on Tuesday, Sept 22, 2026 from 10:00-15:00; tentative schedule, and location is Snijderszaal (LB01.010) at Building 36, TU Delft campus, Mekelweg 4; 2628 CD Delft; register here.

Schedule

10:00: Workshop introduction

10:15: Session by Abigail Nieves, Associate professor at the Freudenthal Institute, Utrecht University, will help us to come to a better understanding of our view on the why, what, and how of metascience for Machine Learning.

12:40: Lunch

13:30: Talk by fundamental machine learning researcher David Tax, Delft University of Technology, with an Analysis of Results in the Machine Learning Research Field.

14:15: Closing.

Speaker presentations:

Session by Abigail Nieves.

In this interactive session, Abigail will introduce the history and philosophy of science followed by a group activity to discover the who, why, and what of Metascience for Machine Learning. If you have an opinion on what the challenges are in the field and why, how to make change etc, this is activity will formally facilitate that process through group discussion and identify concrete effective action.

Invited talk by David Tax.

Research into AI and Deep models / Machine Learning is growing rapidly, which puts strain on the usual scientific checks and reviews. As with any other scientific field, we want to start our investigation by measuring how relevant the issue is, and collect data which will help us analyze it. Here we focus on papers from the largest AI conference NeurIPS, and their own checklist which should correlate with the quality of a scientific paper. We are interested in seeing whether doing these checks using LLMs is trustworthy, and whether there are specific parts of the checklist the LLMs are bad at interpreting.This presentation reports on our first experience of investigating this topic with a BSc thesis project group in Q4 of the last academic year.