Canada CIFAR AI Chair
Pan-Canadian AI Strategy
Guillaume Lajoie is a Canada CIFAR AI Chair at Mila and an assistant professor in the department of mathematics and statistics at Université de Montréal.
Lajoie’s research focuses on guide dynamics in neural networks. Rooted at the intersection of Applied Mathematics, Artificial Intelligence (AI) and Neuroscience, his work seeks to extract lessons from the brain to build artificial systems that learn better, and in turn, develop tools to interact with and uncover brain functions. By pursuing a range of complementary research topics, his goal is to establish synergetic advancements in the following two research axes: 1) biologically-inspired architectural inductive biases in shaping learning and dynamics in artificial neural networks; 2) development of AI tools for brain-machine interfacing with the goals of studying how the brain learns to inform AI, and to allow direct interaction between artificial and biological neural networks for clinical use.
- Google Research Faculty Award, 2020
- Research Scholar award from the Fonds de Recherche du Québec en Santé (FRQS), 2018
- Washington Research Foundation Innovation Fellow, 2015
- Bernstein Fellow in Computational Neuroscience, 2014
Pezeshki, M., Kaba, S. O., Bengio, Y., Courville, A., Precup, D., & Lajoie, G. (2020). Gradient Starvation: A Learning Proclivity in Neural Networks. arXiv preprint arXiv:2011.09468.
Kerg, G., Kanuparthi, B., Goyal, A., Goyette, K., Bengio, Y., & Lajoie, G. (2020). Attention: untangling tradeoffs in self-attentive neural networks, Second Symposium on Machine Learning and Dynamical Systems. arXiv preprint arXiv:2006.09471.
Laferrière, S., Bonizzato, M., Côté, S. L., Dancause, N., & Lajoie, G. (2020). Hierarchical Bayesian optimization of spatiotemporal neurostimulations for targeted motor outputs. IEEE Transactions on Neural Systems and Rehabilitation Engineering.
Kerg, G., Goyette, K., Touzel, M. P., Gidel, G., Vorontsov, E., Bengio, Y., & Lajoie, G. (2019). Non-normal Recurrent Neural Network (nnRNN): learning long time dependencies while improving expressivity with transient dynamics. In Advances in Neural Information Processing Systems (pp. 13613-13623).
Lajoie, G., Krouchev, N. I., Kalaska, J. F., Fairhall, A. L., & Fetz, E. E. (2017). Correlation-based model of artificially induced plasticity in motor cortex by a bidirectional brain-computer interface. PLoS computational biology, 13(2), e1005343.
CIFAR is a registered charitable organization supported by the governments of Canada, Alberta and Quebec, as well as foundations, individuals, corporations and Canadian and international partner organizations.