Mila’s expertise highlighted in ICLR 2024 workshops

Pictures of Mila researchers at the ICLR 2024 event

On May 11, 2024, Mila researchers co-organized and participated in a wide variety of workshops on the margins of the main ICLR 2024 conference in Vienna, tackling societal issues ranging from climate change to Artificial general intelligence (AGI) to underrepresented languages.

This year, Bonaventure Dossou, PhD student at Mila and McGill, opened the 5th Workshop on African Natural Language Processing (AfricaNLP 2024), which he co-organized.

The workshop aimed to build a community and raise awareness around the issue of underrepresentation of African languages in AI research, especially amidst an increased interest in generative AI, Bonaventure Dossou explained.

“Generative AI models have mostly been pre-trained on data in English… We need to adapt this technology to the African context and make sure that we exploit its full potential because at the end of the day, those technologies are built into research, into products that impact people's lives.”

“We hope to increase participation and access of students to the workshop to establish collaborations, share knowledge, and share some cultural aspects and values of the community that we have been building,” he said.

This year, participants “pushed a bit further the AfricaNLP research space through several dataset contributions”, he concluded.

A list of papers accepted at the AfricaNLP workshop can be found here.

 

Mila researchers also co-organized the following workshops:

Tackling Climate Change with Machine Learning: Fostering the Maturity of ML Applications for Climate Change, co-organized by Arthur OuaknineDavid RolnickYoshua Bengio.

Poster: 

Predicting Species Occurrence Patterns from Partial Observations
Hager Radi Abdelwahed, Mélisande Teng, David Rolnick

Machine Learning for Remote Sensing (ML4RS), co-organized by Gabriel Tseng

Poster:

Towards Climate Variable Prediction with Conditioned Spatio-Temporal Normalizing Flows 
Christina Winkler, David Rolnick

 

Here is a list of workshops that featured work by Mila researchers:

How Far Are We From AGI

Keynote talk and panel discussion by Yoshua Bengio.

Poster: 

Efficient Causal Graph Discovery Using Large Language Models
Thomas Jiralerspong, Xiaoyin Chen, Yash More, Vedant Shah, Yoshua Bengio

Workshop on practical ML for limited/low resource settings (PML4LRS)

Contributed talk:

D^2-Sparse: Navigating the low data learning regime with coupled sparse networks
Diganta Misra

Panel discussion:

Parameter and resource efficient development and use of Large Language Models (LLMs)
Bonaventure F. P. Dossou · David Adelani

Data-centric Machine Learning Research (DMLR): Harnessing Momentum for Science

Poster:

GitChameleon: Breaking the version barrier for code generation models
Nizar Islah, Justine Gehring, Diganta Misra, Massimo Caccia, Irina Rish

Workshop on Representational Alignment (Re-Align)

Poster:

Learning and Aligning Structured Random Feature Networks
Vivian White, Muawiz Chaudhary, Guy Wolf, Guilllaume Lajoie, Kameron Decker Harris

Harmony in Diversity: Merging Neural Networks with Canonical Correlation Analysis
Stefan Horoi, Albert Manuel Orozco Camacho, Eugene Belilovsky, Guy Wolf

Generative Models for Decision Making

Poster:

Discrete Probabilistic Inference as Control in Multi-path Environments
Tristan Deleu, Padideh Nouri, Nikolay Malkin, Doina Precup, Yoshua Bengio

Workshop on Mathematical and Empirical Understanding of Foundation Models

Poster:

Predicting Grokking Long Before it Happens: A look into the loss landscape of models which grok
Tikeng Notsawo Pascal Junior, Hattie Zhou, Mohammad Pezeshki, Irina Rish, Guillaume Dumas