ICML

15 new papers from UBC Computer Science at world’s premier machine learning conference

UBC researchers had 15 papers accepted at ICML 2026, including two Spotlight papers 

UBC Computer Science researchers will be in Seoul, South Korea from July 6 - 11, 2026 for the 43rd International Conference on Machine Learning (ICML 2026). ICML is one of the leading conferences for machine learning researchers to gather and discuss the latest advances in the field. 

With nearly 24,000 submissions to the conference this year, a total of 6,352 papers were accepted. This year, researchers from UBC’s Computer Science department had 15 papers accepted into the conference. Two noteworthy papers, one from Professor Margo Seltzer and one from Assistant Professor Evan Shelhamer, were named “Spotlight” papers and are in the top 2.2% of paper submissions.  

In addition to the 15 papers, Professor Mark Schmidt will be giving a tutorial entitled, “Is numerical optimization theory irrelevant to machine learning practice in 2026?” Professor Kevin Leyton-Brown serves as a Social Chair for the conference. 

 

The following papers were accepted at ICML 2026:  

  1. Efficient Rashomon Set Approximation for Decision Trees 
    Zakk Heile, Hayden McTavish, Varun Babbar, Margo Seltzer, Cynthia Rudin 
     

  1. Flatland: The Adventures of Gradient Descent with Large Step Sizes 
    Leonardo Galli, Curtis Fox, Wiebke Bartolomaeus, Mark Schmidt, Holger Rauhut 
     

  1. Pluralistic Leaderboards 
    Nika Haghtalab, Ariel Procaccia, Han Shao, Serena Wang, Kunhe Yang 
     

  1. Rashomon Sets of Falling Trees 
    Varun Babbar, Zachery Boner, Margo Seltzer, Cynthia Rudin 
    Spotlight Poster