Innovative Projects Realized

3d density estimation using normalizing flows and its application to 3d reconstruction in cryo-EM

Generative models enable the researchers to address multiple problems spanning from noise removal to generating novel samples with properties of the domain. Generative models are commonly studied for images and in this project the idea will be expanded to 3D structures or volumes. Single-particle cryo-electron microscopy (cryo-EM) is a technique to estimate accurate 3D structures of biological molecules which is used by practitioners in fields like precision medicine. This allows them to design drugs that could cure patients with rare diseases and avoid side effects. A trained generative model on previously estimated molecular density models, would enable rapid improvement in resolution of estimated densities of limited resolution. The outcome of this research project will be provided as a ready-made tool that improves the resolution of estimated densities in its input. Through this collaboration, Borealis AI would push forward 3D generative methods research and its application to density estimation.

Évaluation de la résilience écologique et économique d’agroforesterie du cacao aromatique

Le projet de recherche prend place dans le sud-ouest du Mexique, au Soconusco. Il a pour cadre le Centre Agroécologique Saint François d’Assise (CASFA). Cette coopérative, qui regroupe près de 300 familles d’agriculteurs, a mis en place un modèle de culture d’agroforesterie du cacao, c’est à dire que les cacaotiers sont cultivés sous un couvert forestier. Le cacao est ensuite vendu à des chocolateries fines françaises et belges. La coopérative regroupe des modèles de culture plus ou moins diversifiés (avec plus ou moins d’espèces sur la parcelle agricole). L’objectif est donc d’évaluer la capacité de résilience des différents types de cultures de ce modèle d’agroforesterie, c’est à dire leur capacité d’adaptation à un choc ou à un stress extérieur. Pour cela, des indicateurs de résilience écologique et économique sont développés. TO BE CONT’D

Road mortality in Toronto and region: The utility and need for wildlife passages in a fragmented landscape

The Toronto and Region Conservation Authority (TRCA) jurisdiction is one of the most densely populated watersheds in Canada, where the urban area now accounts for close to half of the land cover in the region. The conversion of land from natural to urban area, such as roads, contributes to the loss and fragmentation of habitat, which is a major driver of biodiversity declines. This project investigates the potential of various passage structures that can mitigate these impacts and assesses the need for strategic wildlife crossings. The scope of the project aims to develop a systematic field method to (i) assess the effectiveness of wildlife crossing structures (e.g., fencing, passages) in TRCA’s jurisdiction and (ii) identify specific locations at pilot sites for mitigation measures to reduce road mortality and improve connectivity. Ultimately the goal is to reduce biodiversity declines, where species can support functioning and healthy ecosystems within the urban environment.

Ecological restoration of natural woodland and meadow sites invaded by dog-strangling vine (Vincetoxicum rossicum) in southern Ontario.

Dog-strangling vine is one of the most invasive plants in eastern Canada. Previous attempts to control the spread of the vine have been unsuccessful. Recent research on this invader suggests that when it invades an area, it perturbs soil microbial communities. These perturbations appear to interfere with some native plants’ ability to grow. However, other native plants appear to be resistant to these soil alterations. This work aims to identify additional plant species that are resistant to the soil effects of dog-strangling vine, to determine how long these soil effects last, and to identify soil amendments that can reduce the impact of invasion. This research will help develop much-needed management and restoration strategies and will help protect and revitalize the fragile ecosystems that dog-strangling vine currently threatens across eastern Canada.

AI Techniques to Explore the Relationship Between Structural and Function Brain Connectivities

It is not fully understood how structural brain connectivities give rise to functional brain connectivities. The objective of this project is to apply Deep Learning and Machine Learning techniques to explore the complex relationship between structural and functional brain connectivities and accurately describe this important structure-function relationship. A linear brain dynamic network model will be learnt, studied developed and analyzed. This model’s prediction capabilities to infer function correlation (obtained from EEG, MEG and fMRI) from structural connectivity (obtained from diffusion MRI) will be tested and validated through experimentation.

Functionalized adsorbent for simultaneous removal of heavy metal ions and ciprofloxacin

Ciprofloxacin (CIP) is a broad-spectrum antibiotic, frequently detected in different environmental compartments, which may contribute to antibiotic resistance occurrence. Another group of the contaminants, widely present in water and soils is heavy metal ions (HMI), like mercury, lead or cadmium. They are known to have toxic or poisonous effect on living organisms, including humans, even at low concentration. Moreover, CIP tends to form stable complexes with metal ions, including HMI. This may alter antibiotic’s antimicrobial and physicochemical properties. The objective of this project is to use the later feature to remove CIP and HMI from water. The aim is to create cost-effective adsorbent, so that CIP and heavy metals are adsorbed onto its surface. The adsorption of metal ions, including HMI would occur in the beginning which is most likely to enhance CIP adsorption. Hence, the decrease of their concentration in aquatic media is expected.

Ships and Whales in Gitga’at Territory: Risks of whale-vessel interactions.

As whales recover from the past centuries of whaling and as global trade compels shipping traffic to increase, we expect negative whale-ship interactions such as fatal strikes and noise disturbance to become an increasingly serious issue. To develop cost-effective and broadly applicable methods of assessing the risks of ships to whale feeding grounds, I propose to work with the partner organization to conduct a vessel strike risk assessment based on 3 years of visual survey data in a remote fjord system of northern British Columbia that is slated for increased shipping traffic. I will analyze the data we collect to assess the risk of strikes with small recreational vessels and large ships for two whales species, humpback whales and fin whale. The methods we develop will inform mitigation measures in our study site and facilitate similar studies elsewhere.

Intelligence artificielle appliquée pour l’analyse, l’optimisation et l’innovation

Ce projet vise à développer et approfondir les connaissances et techniques nécessaires à la mise en oeuvre des algorithmes d’intelligence artificielle sur des données issues du monde réel. En partenariat avec de multiples acteurs industriels de la grande région de Québec, la Faculté de Science et Génie de l’université Laval, au travers de ses étudiants à la maîtrise professionnelle en informatique – Intelligence artificielle, offre son expertise en recherche et développement pour encadrer la recherche et le développement nécessaire au déploiement d’algorithmes issus de l’intelligence artificielle aux nombreuses problématiques rencontrés par lesdits partenaires de ce projet. Par l’analyse automatique de flux vidéos, de texte, de données 3D, ou géoréférencées, et jusqu’à la la production de prédictions et de décisions pertinentes au domaine des partenaires, les étudiants participants à ce projet favoriseront l’innovation et la compétitivité de nos partenaires industriels.

Multi-Document, Aspect-Based Sentiment Analysis of Political News Articles

The main objective of the project is to upgrade the existing system at Gnowit to create a complex self learning automated decision support system. The system(crawlers) will automatically collect all the data related to any company, political system or a specific entity from the articles, blogs and reviews and create a more actionable report. The system will also check the credibility of the news and perform the controversy analysis. The mined information and knowledge will be saved and used to train a engine which could predict the future events of an entity or object. The system also tries to quantify the sentiment between multiple aspects.

Towards Automation of 3D Visualization & Analysis of Workspace Collisions on Construction Jobsites

Workspace Interferences/Collisions happen on construction jobsites when multiple resources (labor, equipment, material, etc.) don’t have enough space to coexist at the same time and will interfere with each other’s operations. These interferences affect performance, delaying the project, impacting cost, and may jeopardize the buildings’ integrity and people’s safety on site. Most of existing models simulate resources as being deterministic. However, the behavior of the crew interfering is more chaotic. The chaos mainly emerges from the dependency of the interferences and their consequences on the human interactions among crew members and between different crews. 85% of the respondents of a published questionnaire entailed the necessity for modelling the propagating effects of interferences and considering workspaces as stochastic. Accordingly, the research for this internship will focus on: 1) identifying the sources of uncertainty in workspace modelling 2) Testing the sensitivity of existing workspace models towards discovered uncertainties 3) Modifying the existing evaluation methods to account for uncertainties

Pure-sine GaN-based motor inverter

This project applies wide-bandgap (WBG) transistors to voltage level multiplier module (VLMM) topology in motor inverter applications. It is expected that this approach can yield the benefits of WBG motor inverters (high motor efficiency, fast control response, lower motor torque ripple, close to ideal sinusoidal motor current waveform, smaller filter size, lower cost filter, etc.) while leveraging the benefits of VLMM (lower component cost, high frequency switching only at low voltage, filter-less output signal) to yield a commercially viable highly-efficient pure-sine motor inverter.

Quantifying magnetic susceptibility of the human brain with MRI