Innovative Projects Realized

Programmable Transport in Hardware

As network speeds reach hundreds of gigabits per second and beyond, there is increasing interest in using programmable Network Interface Cards (NICs) to improve network performance. A key area of focus is accelerating transport protocols, which significantly impact network performance. Transport protocols are continuously re-designed and updated, especially in large-scale networks, to optimize network performance. However, programming these complex protocols on NICs in an efficient manner is difficult and error-prone, requiring deep knowledge of low-level programming on NIC hardware. This project aims to solve this problem by developing a platform that allows users to write transport protocols in a high-level programming language, without needing to understand the hardware intricacies of NICs. A compiler will translate these protocols into optimized code for the NIC hardware. AMD is a leader in high-performance computing, and with its acquisitions of Xilinx and Pensando, both experienced in programmable NICs, is well-positioned to shape the future of this field. This project will allow AMD’s customers to accelerate transport protocols more easily, reducing the risk of human errors in protocol implementation, potentially avoiding network disruptions and ensuring better performance for millions of users.

Développent d’outils innovateurs pour la caractérisation du bois et de la fibre du bois et leur impact sur la rentabilité financière des travaux d’aménagement forestier intensif

L’aménagement forestier intensif est parmi les alternatives pour assurer la pérennité des ressources pour l’industrie forestière. Cependant, peu d’information est disponible sur les effets de ces pratiques sur la qualité du bois et la valeur des produits. Ce manque d’information est dû à la nature destructive des essais de caractérisation des propriétés du bois. Ainsi, ce projet vise à développer des outils rapides et non destructifs pour mesurer les implications des pratiques d’aménagement forestier intensif sur les indicateurs de la qualité du bois et sur la valeur des produits. De plus, les modèles de rentabilité financière et d’évaluation économique existants n’intègrent pas la qualité du bois et ses répercussions sur la valeur des produits. En ajoutant ces paramètres aux modèles de rentabilité financière, ce projet vise aussi à apporter une estimation plus réaliste des coûts/bénéfices des travaux d’aménagements forestiers intensifs et des produits issus de la transformation. Le projet repose sur des données empiriques relatives aux propriétés chimiques et mécaniques du bois et à l’évolution des volumes marchands et de la dimension des tiges dans soumises à l’éclaircie commerciale et à la fertilisation pour une période de 14 ans après traitements.

Assisted Video Laryngoscopy (AVL): A Novel Method of Endotracheal Intubation Using a Next-Generation Video Laryngoscope with Artificial Intelligence Assistance

The research project aims to improve the safety and effectiveness of endotracheal intubation, a critical medical procedure. Endotracheal intubation involves inserting a breathing tube into a patient’s airway using a laryngoscope and is commonly required for surgery and in emergency situations. The research intern will study a novel technique called assisted video laryngoscopy (AVL), achieved by incorporating artificial intelligence into a laryngoscope. Several challenges that medical professionals encounter during intubation will be addressed, including limited visibility, inadequate training, and ergonomic issues with existing devices.
The partner organization will obtain an innovative laryngoscope that could significantly improve the first pass success rate of endotracheal intubation and reduce the risk of complications. This could lead to better patient outcomes and save lives, especially in time-critical situations where successful intubation is crucial to prevent cerebral hypoxia and its devastating consequences, such as brain damage or patient fatality.

Catégorification des partitions non croisées généralisées

Le domaine du stage proposé est à l’interface entre la combinatoire et l’algèbre. Plus spécifiquement, il s’intéresse à la combinatoire apparaissant en théorie des représentations d’algèbres. Une question centrale dans ce domaine est la catégorification. Le principe est, étant donné un phénomène combinatoire, de trouver/construire une catégorie (typiquement une catégorie provenant de la théorie des représentations d’algèbres) dans laquelle la combinatoire étudiée apparaît naturellement. Cela permet notamment de mieux comprendre le phénomène combinatoire, mais aussi d’en déduire des propriétés intéressantes sur la catégorie étudiée.

L’objet combinatoire étudié dans ce projet est la notion de partitions non croisées sur une surface à points marqués. L’objectif principal du stage est le suivant : associer à une surface à points marqués, une certaine catégorie, et un certain sous-ensemble de sous-catégories formant un treillis isomorphe au treillis des partitions non croisées de cette surface. Ceci généraliserait largement des résultats déjà existant si la surface est un disque : la catégorie utilisée est alors celle des représentations d’un carquois de type A de Dynkin.

Ce projet permettra de mettre en commun l’expertise en théorie des représentations associées à des surfaces des superviseurs français, et celle en combinatoire algébrique du superviseur canadien et des membres de son laboratoire.

Fluent Interactions: Comparing conversations in online and face-to-face L2 French language learning classrooms to see if there is an impact on the motivation of learners

Comparing virtual and face-to-face interactions in language learning classrooms, this study aims to understand how the smoothness of turn-taking, interaction quality, and the presence of visual cues (such as body language) influence adult learners’ motivation to continue studying French in Quebec. We intend to conduct this through a mixed methods approach. Controlled experiments manipulating turn-taking conditions and visual cues will assess their effect on communicative reward and motivation of L2 French learners (generating quantitative data). Following these experiments we will have discussions with the participants about what factors make L2 language learning rewarding in each setting. This will provide qualitative insights into L2 learning motivation factors. This project is designed to replicate our current study which is conducting these experiments in the context of L2 Welsh adult learners. Comparing the findings of both studies will allow us to examine whether the findings are consistent cross-culturally and develop effective pedagogical strategies for policymakers and educators to sustain language learning motivation in adult learners globally. This study will facilitate a collaborative link between McGill and Cardiff Metropolitan University, leveraging the unique suitability of Welsh and Canadian contexts for comparative language education research and fostering a specialised cross-cultural knowledge exchange.

Stage de recherche sur les techniques des retombées atmosphériques appliquées à l’érosion du sol et la sédimentation

Le stage de recherche vise à approfondir la compréhension par le stagiaire des principes de bases et de l’application des méthodes utilisant les radionucléides de retombées atmosphériques pour reconstruire les processus d’érosion et de sédimentation. Le stagiaire bénéficiera de formations techniques en spectrométrie gamma, et de soutien pour mettre en œuvre les modèles de calcul de redistribution de sol à partir des inventaires de radionucléides, mais aussi pour l’établissement de modèle d’âge pour les archives sédimentaires à partir de mesures de profils de radionucléides.

Cela lui permet d’avoir des compétences et expertises précieuses en matière de gestion des sols face à leur dégradation à l’échelle des bassins versants, notamment dans le contexte du changement climatique.

Landscapes of Minority Language Learning

Both Wales and Quebec are known for the importance they place on promoting the Welsh and French languages, respectively, and have introduced a variety of policy, cultural, and educational initiatives to this end. They owe these successes partly to their history of collaborating and sharing knowledge to enhance language revitalization and education efforts in both contexts. This project will build on this history of collaboration. It will document adult newcomers’ linguistic integration trajectories through formal learning, such as community language classes, as well as informal learning such as seeking out opportunities to speak the minority language during their daily tasks and joining minority-language social groups. This research will benefit both Canadian and Welsh contexts by providing a more complete view of linguistic integration than is currently available. It will also highlight the agency and resourcefulness of adult language learners and will reveal the barriers that hinder their linguistic integration, opening up new opportunities to better support these learners’ full inclusion and participation in their new communities.

Artificial Intelligence through machine learning to explore individual signatures in animal sounds

Our team investigates vocalizations of Savannah Sparrows (Passerculus sandwichensis) to understand how vocalizations are used during the breeding season. Our team previously detected a distinctive call type – the “twitter call”- that was poorly described and understood. The function of this call is unknown. The goal of this project is to explore if the twitter call contains an individual signature allowing birds to identify each other by call alone. Using recordings previously collected in the field, we will work with Dr. Nicolas Mathevon of the University of Saint-Etienne and ENES Bioacoustics Research Lab to use Artificial Intelligence to apply a new method developed by his team to detect individuality in the calls of individual birds. These methods allow a more accurate analysis of calls and increase our ability to detect the specific acoustic features that provide individuality. Knowing if this call is individually specific will increase our understanding of possible functions of this call and direct future experiments to test these functions. This research will additionally benefit Dr. Mathevon and his team by providing a full dataset to continue testing their new methodology, which has been previously used to explore more complex bird songs, rather than simple calls.

Effective Prediction and Recommendation by Concentrating on Dynamicity and Evolution in Charity Data Processing

Individuals (donors) wishing to donate funds to non-profit and/or charitable organizations want to know how the organizations operate and how effectively their donations are used. Dexterity has developed an online service called Place2Give which aims to answer those questions. The research project aims to employ social network analysis and data mining to enhance the data introduced into the databases used by Place2Give from which the data for the analysis of the charitable organizations behavior is extracted on a case by case basis.

Aggregation of Distributed Residential Energy Storage for Enhanced Grid Resilience and Renewable Energy Growth

THIS IS A GENERIC TEXT PUT IN PLACE AS THERE WAS NO PROJECT OVERVIEW

Characterisation of binder materials in silicon-graphite electrodes from slurry to lithium-ion battery

The scaling up of lithium-ion battery technology for high energy applications such as powering electric vehicles is not without its challenges. Increasing the capacity of these large energy storage systems is key in addressing consumer concerns about the driving range of electric vehicles, which remains inferior to that of gasoline-powered vehicles on average. The use of new materials within the battery electrodes is a promising avenue to doing just that. As such, the present study focuses on silicon-graphite composite electrodes, which have a significantly higher capacity per unit weight than standard graphite electrodes. However, issues with maintaining this capacity over many uses persist. One way to stabilize these electrodes is to optimize the binder that ensures the cohesion of the electrode material. The present study focuses on polyacrylic acid-based polymers and the effect of modifying several key parameters such as the solution pH, the water content, and the chain length of the polymer on the electrochemical performances of the electrodes. The information gained from this study will help both research groups improve their battery formulations going forward and foster collaboration between Canadian and French universities and industry partners.

Developing predictive models for air pollution and heat exposures in South Africa and Canada

The overall objective of the Mitacs Globalink Project is to develop and assess the validity of models of exposure to air pollutants and temperature variability, using data extracted from a variety of available sources in South Africa and Canada. This objective will be achieved through a series of specific aims.

The first of these aims will be (1) the identification of the sources of available datasets, using established networks in South Africa and Canada to locate, access and extract the relevant data from these sources. Ground station data in both countries is generally available through publicly accessible portals.

Aim (2) will require subjecting these datasets to rigorous and systematic evaluation to determine their quality. This will require using established methods of data checking, and a review of the quality control methods used in data collection. Having determined the quality of the data, the next step is ensuring validity and ground truthing. This will require a variety of methods of data triangulation.

The data will then be used to achieve Aim (3), to use the validated datasets to develop predictive models using machine learning (ML) methods for air pollution and heat exposures using satellite earth observations and ground-level monitoring stations.