Hello! Bonjour! سلام!
I’m Abbas Mehrabian, a mathematician, freelance journalist, and Google DeepMind research scientist living in Montreal.
These activities instill meaning in my life: creating beautiful content, learning new skills, pushing my limits, building connections with others, and spending time in nature.
To discover more, read my 8-page personal essay (updated in December 2023) or watch the first 25 minutes of my talk (from January 2022).
Find me on social media or email me at abbas dot mehrabian at gmail dot com.
Biography
I was born in Pakistan in 1986, raised in Iran, and moved to Canada in 2009.
Since June 2021, I am a research scientist in artificial intelligence at Google DeepMind’s Montreal team led by Doina Precup.
With a PhD in mathematics and a graduate diploma in journalism, I have years of experience in academic research in mathematics and computer science, and I have done journalism in both English and Persian.
I love travelling, and I had lived in Tehran, Waterloo (Ontario), Melbourne, Toronto, Vancouver, and Berkeley (California) before coming to Montreal in 2017.
My hobbies include camping, cycling, reading, and photography.
Journalism
At age 33, I started doing journalism because I wanted to learn more about Canadian society and have a broader impact.
So, I completed a graduate diploma in journalism at Concordia University, interned at Broadview magazine for four months, and I’m now a freelance journalist and a member of
Quebec Writers' Federation,
the Canadian Association of Journalists, and
the Society of Professional Journalists.
At age 36, I started a Persian podcast,
این قصه منه (This is My Story).
I’m interested in writing human interest stories and, in particular, social justice stories. See my English publications, English essays, and Persian publications.
Mathematics and computer science
I received a PhD in mathematics in 2015, and I have worked in areas such as probability theory, randomized algorithms, random matrices, random graphs, and graph theory.
My focus has been on the theoretical aspects of machine learning since 2016. Two of my best publications were:
-
The sample complexity of learning Gaussian mixtures, published in the Journal of the Association for Computing Machinery (JACM) and won a best-paper award at the Conference on Neural Information Processing Systems (NIPS 2018), and
VC-dimensions of neural networks, published in the Conference on Learning Theory (COLT 2017) and the Journal of Machine Learning Research (JMLR).
Check out my research projects and research publications (or DBLP or Google Scholar).
Awards
1. Governor General’s Gold Medal for the best PhD thesis at University of Waterloo, 2015
2. IVADO Postdoctoral Fellowship, 2018–2020 ($90,000 per year)
3. Best-Paper Award at 2018 Conference on Neural Information Processing Systems, out of 4856 submissions
4. Sportsnet Diploma Scholarship in Journalism for the personal essay Love Your Neighbour as Yourself, 2021
