There has been an emerging interest in edge implementation of deep neural networks, but this direction has no specific scientific event dedicated to it, and this workshop aimed at filling this gap. Edge intelligence is a highly promising area in AI, which is identified as one of the top 10 breakthrough technologies (also known as TinyAI) in 2020 by MIT technology review. This workshop was the first major event in Canada dedicated to this topic.

The main purpose of the workshop was to bring attention to challenges of running artificial intelligence algorithms on edge devices, with constrained hardware. The areas covered in the workshop include but not limited to computer vision, natural language processing, speech processing, theory of machine learning, statistics, deep learning, statistics, graph convolution network, efficient computing, cloud computing, hardware, software, optimization, neural model compression, data compression, among others.

The workshop was co-chaired by Prof. Andrea Lodi, a member of Group for Research in Decision Analysis (GERAD) and the holder of the Canada excellence research chair in data science for real-time decision making (DS4DM) and Dr. Vahid Partovi Nia a GERAD a Principal Researcher of Huawei Noah’s Ark Lab and a GERAD associate member. The scientific organizing committee were composed of the co-chairs, Prof. Sébastien Le Digabel from GERAD, and Dr. Mehdi Rezagholizadeh from Huawei Noah’s Ark Lab.

The Edge Intelligence Workshop hosted more than 20 technical talks given by professors from top Canadian universities including Polytechnique Montréal, Université de Montréal, McGill, INRS, University of Toronto, University of Waterloo, and from top Canadian research institutes such as, Centre Research Mathématiques (CRM), Montréal Institute for Learning Algorithm (MILA). A keynote talk was given by Prof. Yoshua Bengio on the energy efficient deep learning, along with an inspiring talk by Prof. Andrea Lodi on connection between the linear integer programming with artificial intelligence algorithms. Several technical talks were given by researchers of Huawei’s Noah’s Ark Lab to cover the industry vision on the topic.

More than 20 refereed papers were accepted for poster presentation which will appear in the proceedings of the workshop. A paper from University of Toronto electrical and computer engineering department and a paper from University of Waterloo computer science department won the first and the second best paper prize, respectively.

This event was a successful platform to establish relationship and exchange of ideas between the industry and the academia through technical talks and poster discussions. GERAD, CRM, and DS4DM were the main academic sponsors of the event and Huawei Noah’s Ark Lab was the main industrial sponsor.

The organizing committee thank the participants, invited speakers, CRM, and DS4DM, especially GERAD employees Marie Perreault, Marilyne Lavoie, and Karine Hébert for making this event a successful practice. GERAD hosted this event at HEC Montréal and the organizing committee preview a similar event in 2021.