Simon Guiroy
simonguiroy at umontreal dot ca

I am a PhD candidate in Deep Learning at MILA, enrolled at the Universite de Montreal. I am supervised by Prof. Chris Pal, and also co-supervised by Prof. Sarath Chandar . I completed my Master's in Deep Learning (Computer Science) at MILA in 2019, and completed my Bachelors of Electrical Engineering at Polytechnique Montreal in 2016.

I've studied a variety of graduate courses and topics in machine learning, including Deep Learning, Reinforcement Learning, Probabilistics Graphical Models, Continual Learning, Computer Vision.

My non differentiable passions include music, playing flamenco guitar and learning foreign languages.

Email  /  CV  /  LinkedIn  /  Github  /  Twitter

profile photo
Research

My main research interests are Deep Learning and the wide field of "Learning to Learn". I am interested in understanding how deep neural networks can learn representations that generalize well to new data distributions, tasks and domains, and how neural networks can "learn to learn" such representations.

Teaching experience
Papers
3DSP Towards Understanding Generalization in Gradient-Based Meta-Learning
Simon Guiroy, Vikas Verma, Christopher Pal
ICML 2019 Workshop, Understanding and Improving Generalization in Deep Learning
[paper]
3DSP On the reproducibility of gradient-based Meta-Reinforcement Learning baselines
Tristan Deleu, Simon Guiroy, Seyedarian Hosseini
ICML 2018 Workshop, Reproducibility in Machine Learning
[paper]
3DSP Application of the Kaldi toolkit for continuous speech recognition using Hidden-Markov Models and Deep Neural Networks
Simon Guiroy, Ricardo de Cordoba, Amelia Villegas
Iberspeech, 2016
[paper]
Projects
3DSP Recognizing Violent Human Actions in Video with Deep Learning [GitHub]
3DSP Extracting Information using Ocular Commands and Augmented Reality [Report]

(Design and CSS courtesy: Jon Barron and Amlaan Bhoi)