IML Machine Learning Working Group

Tuesday 15 Nov 2022, 15:00 → 18:00 Europe/Zurich

13/2-005 (CERN)

13/2-005

CERN

90

Show room on map

Description

Topic: Model interpretability

96543252431

Simon Akar

Riccardo Torre, Fabio Catalano

- 15:00 → 15:05
  
  News 5m
  
  Speakers: Anja Butter, Fabio Catalano (University and INFN Torino (IT)), Lorenzo Moneta (CERN), Michael Kagan (SLAC National Accelerator Laboratory (US)), Dr Pietro Vischia (Universite Catholique de Louvain (UCL) (BE)), Simon Akar (University of Cincinnati (US)), Stefano Carrazza (CERN)
  
  Seminar on relating uncertainties in ML and HEP (PHYSTAT + CERN DS)
- 15:05 → 15:30
  
  Attribution-based Explanations that Provide Recourse Cannot be Robust 25m
  
  Speaker: Prof. Rianne de Heide (Vrije Universiteit Amsterdam)
  
  arxiv:2205.15834
  
  Slides
- 15:30 → 15:35
  
  Question time 5m
- 15:35 → 16:00
  
  A Detailed Study of Interpretability of Deep Neural Network based Top Taggers 25m
  
  Speaker: Avik Roy (University of Illinois at Urbana-Champaign)
  
  2022-11-15_TopTaggerInterp.pdf
  
  arxiv 2210.04371
- 16:00 → 16:05
  
  Question time 5m
- 16:05 → 16:30
  
  Interpretable and Trustworthy Geometric/Graph Deep Learning. 25m
  
  Speaker: Siqi Miao (Purdue University)
  
  arXiv:2201.12987
  
  arXiv:2210.16966
  
  LRI-IML.pdf
- 16:30 → 16:35
  
  Question time 5m
- 16:35 → 17:00
  
  (Postponed or virtual) Summary of the Second MODE Workshop 25m
  
  Second MODE Workshop on Differentiable Programming for Experiment Design
  
  https://indico.cern.ch/event/1145124/
  
  Speaker: Dr Pietro Vischia (Universite Catholique de Louvain (UCL) (BE))
  
  Workshop agenda
- 17:00 → 17:05
  
  Question time 5m