Simon Lacoste-Julien

Simon Lacoste-Julien is a machine-learning researcher whose work spans Bayesian inference, structured prediction and optimization. He earned a triple-honours BSc in mathematics, physics and computer science at McGill in 2003 and a PhD in computer science and statistics at UC Berkeley in 2009, then completed postdoctoral work at Cambridge and returned to Montréal in 2016 after research at INRIA and École normale supérieure in Paris. He coauthored papers on loss-calibrated Bayesian inference and block-coordinate Frank-Wolfe optimization, then joined Samsung's SAIT AI Lab Montréal as its leader in 2019. As of October 2026, he is Associate Scientific Director at Mila and a full professor in Université de Montréal's Department of Computer Science and Operations Research.

Now at: Mila and Université de Montréal
Works: 171
Citations: 6,724
h-index: 35
Badges: Bridge

Publication numbers from OpenAlex, as of 27 October 2026.

Timeline

2003–2009

McGill and Berkeley degrees

He earns a triple-honours BSc in mathematics, physics and computer science at McGill in 2003. In 2009, he completes a PhD in computer science and statistics at UC Berkeley under Michael I. Jordan.
2011

Loss-calibrated Bayesian inference

In 2011, he coauthors “Approximate Inference for the Loss-Calibrated Bayesian” with Ferenc Huszár and Zoubin Ghahramani. The paper proposes an EM-like method that calibrates approximate inference to the decision task and improves Gaussian-process classification when losses are asymmetric.
2013

Block-coordinate Frank-Wolfe

At ICML in 2013, he coauthors “Block-Coordinate Frank-Wolfe Optimization for Structural SVMs” with Martin Jaggi, Mark Schmidt and Patrick Pletscher. The paper proposes a randomized block-coordinate Frank-Wolfe method for block-separable convex optimization and applies it to the dual structural SVM objective.
2016

Return to Montréal

After a postdoc at the University of Cambridge, he works for several years as a researcher at INRIA and École normale supérieure in Paris. In 2016, he returns to Montréal to help grow the city's AI ecosystem.
2019

Samsung SAIT lab leadership

In 2019, he joins Samsung's SAIT AI Lab Montréal as its lab leader. The lab focuses on unsupervised learning and GAN research for deep-learning algorithms and on-device AI.
As of October 2026

Mila and UdeM appointments

As of October 2026, he is Associate Scientific Director at Mila and a full professor in Université de Montréal's Department of Computer Science and Operations Research.