Making Vancouver proud: 11 UBC Computer Science papers accepted for NeurIPS conference

11 papers from the UBC Computer Science Department were accepted for this year's conference on Neural Information Processing Systems.

Once again, the department’s professors and students have an exceptional showing at NeurIPS 2024, with three Spotlight papers (the top 2.08% of papers at the conference qualify as Spotlights).

This year the authors have another great reason to be excited: location, location, location. That’s because the conference is being hosted in UBC’ s hometown: beautiful Vancouver, BC, Canada, at the Vancouver Convention Centre from Tuesday, Dec. 10 through Sunday, Dec. 15. 

NeurIPS is one of the most prestigious conferences in the fields of Artificial Intelligence, Machine Learning, and Data Science. It serves as a global platform for researchers and industry leaders to present their findings, share ideas, and discuss the future of these rapidly evolving fields. The University of British Columbia’s Computer Science Department has had a strong showing multiple years in a row, with 11 papers accepted for the 2023 conference and an impressive 13 papers accepted for the 2022 conference.

Below is a list of all eleven papers accepted for this year’s conference:
 

1. The Edge-of-Reach Problem in Offline Model-Based Reinforcement  Learning
   Anya Sims, Cong Lu, Jakob Foerster, Yee Whye Teh 
    
2. Pre-trained Text-to-Image Diffusion Models Are Versatile Representation Learners for Control 
   **Accepted as a Spotlight paper**
   Gunshi Gupta, Karmesh Yadav, Yarin Gal, Dhruv Batra, Zsolt Kira, Cong Lu, Tim G. J. Rudner
    
3. Bias Amplification in Language Model Evolution: An Iterated Learning Perspective  
   Yi Ren, Shangmin Guo, Linlu Qiu, Bailin Wang, Danica J. Sutherland
    
4. 3D Gaussian Splatting as Markov Chain Monte Carlo  
   **Accepted as a Spotlight paper** 
   Shakiba Kheradmand, Daniel Rebain, Gopal Sharma, Weiwei Sun, Jeff Tseng, Hossam Isack, Abhishek Kar, Andrea Tagliasacchi, Kwang Moo Yi
    
5. Even Sparser Graph Transformers 
   Hamed Shirzad (+ collaborators from Google and CMU), Danica Sutherland  
    
6.  Adaptive Randomized Smoothing: Certified Adversarial Robustness for Multi-Step Defences
   **Accepted as a Spotlight paper** 
   Saiyue Lyu, Shadab Shaikh, Frederick Shpilevskiy, Evan Shelhamer (currently at Google DeepMind but joining UBC CS in 2025), Mathias Lécuyer
    
7. PANORAMIA: Privacy Auditing of Machine Learning Models without Retraining  
   Mishaal Kazmi, Hadrien Lautraite, Alireza Akbari, Qiaoyue Tang, Mauricio Soroco, Tao Wang, Sébastien Gambs, Mathias Lécuyer
    
8. First-Explore, then Exploit: Meta-Learning Intelligent Exploration
   Ben Norman, Jeff Clune
    
9. Heavy-Tailed Class Imbalance and Why Adam Outperforms Gradient Descent on Language Models. 
   **Accepted as a Spotlight paper**
   Frederick Kunstner, Robin Yadav, Alan Milligan, Mark Schmidt, Alberto Bietti
    
10.  Extending Video Masked Autoencoders to 128 frames  
    Nitesh Bharadwaj Gundavarapu, Luke Friedman, Raghav Goyal, Chaitra Hegde, Eirikur Agustsson, Sagar Waghmare, Mikhail Sirotenko, Ming-Hsuan Yang, Tobias Weyand, Boqing Gong, Leonid Sigal
     
11. Interpretable Generalized Additive Models for Datasets with Missing Values
    Hayden McTavish, Jon Donnelly, Margo Seltzer, Cynthia Rudin

In addition, Dr. Evan Shelhamer, who is joining UBC as a faculty member in 2025, Kwang Moo Yi, Mark Schmidt, and Leonid Sigal are all serving as Area Chairs for NeurIPS 2024. This is a senior program committee role responsible for managing peer review and advising the acceptance/rejection of papers for the conference.

The department is extremely proud all these UBC Computer Science faculty and students who put so much hard work into advancing AI and ML with their research.