Innovative Projects Realized

Insights and Recommendation Extraction using advances in Large Language Modeling

Businesses allocate a significant amount of financial resources and time to produce regular financial and managerial reports,
which involve analyzing accounting data. These reports play a vital role in evaluating and comprehending the overall
performance of the business, thereby aiding in making important decisions. Additionally, businesses often engage industry
experts or consultants to derive actionable recommendations and strategies based on the findings from the analysis. The
ultimate goal is to deliver these recommendations in a clear and easily understandable format, typically using a SaaS model
application. Automating this entire process comes with several benefits, such as enhancing operational efficiency, reducing
the likelihood of manual errors, and ultimately leading to cost savings. This project is specifically focused on extracting and
generating meaningful business recommendations using interpretable Large Language Models (LLMs). These LLMs are
sophisticated language models that can process and understand vast amounts of data. Ensuring the accuracy and
effectiveness of these generated recommendations is critical to maintaining the quality of the automatically generated
reports. The project aims to revolutionize the way businesses perform analysis and make decisions. It has the potential to
transform the current landscape of analysis and decision-making processes, making them more data-driven, efficient, and
precise.

Towards a theory of territorial justice

The research project aims to study how the current dynamics of spatial production determine the pattern of global inequality, how the urban “way of life” is shaping the appropriation of natural spaces and what elements can configure a theory of justice that contributes to a positive redefinition of territorial policies. Faced with a context and, above all, the perspective of a future dominated by logics that produce and reproduce territorial inequalities, the central question that this research seeks to explore is what it means to build just cities in the context of planetary urbanization. To this objective, it builds on the expansion of the traditional idea of urbanization as the growth of cities towards a more complex and holistic concept as the process of territorial restructuring.

Evaluating Learning Rules in Visual Perceptual Learning Using Deep Neural Networks

This research project aims to explore and outline the implications of supervised and reinforcement learning dynamics in Deep Neural Networks (DNNs) with regard to Visual Perceptual Learning (VPL). DNNs, which are hierarchical computational models inspired by the biological brain, will be used to simulate various VPL tasks. The focus is on assessing how these distinct learning approaches—supervised, where networks are explicitly trained with correct responses, and reinforcement, where learning occurs via trial and error feedback—affect task performance and learning transferability. These effects will be evaluated by comparing the DNN outcomes with known human and animal perceptual learning characteristics The ultimate goal is to enhance our understanding of the underlying mechanisms of perceptual learning and to determine which learning paradigm most accurately replicates biological processes. This study promises to contribute significantly to our theoretical understanding of VPL and provide valuable insights into the design of more efficient, biologically-inspired artificial learning systems.

Conception d’un capteur microfluidique 3D à résonateur micro-ondes pour les liquides

Le présent stage de recherche a pour objectif d’explorer la conception et l’étude d’un capteur microfluidique à résonateur micro-ondes dans le but de caractériser et de détecter des liquides. Ce stage s’inscrit dans le cadre d’un projet de recherche doctoral mené par le stagiaire à l’Université du Québec à Trois-Rivières (UQTR), en collaboration avec une équipe de recherche spécialisée dans les technologies de micro-ondes de l’Université Grenoble Alpes (UGA), en France.

Le stagiaire aura pour mission de développer une nouvelle structure de capteur microfluidique à résonateur micro-ondes en utilisant la technologie CSRR (complementary split-ring resonator) et le mode différentiel. Cette tâche impliquera l’utilisation de logiciels de simulation de micro-ondes tels que HFSS et ADS, ainsi que de nouveaux outils de l’UGA pour concevoir et optimiser le capteur.

La fabrication du capteur sera effectuée en collaboration avec le Centre National Intégré du Manufacturier Intelligent (CNIMI) en utilisant l’imprimante 3D DragonFly. Le stagiaire sera chargé de réaliser des tests expérimentaux et d’évaluer les performances du capteur en utilisant les équipements de l’équipe de recherche de l’UGA, tels que l’analyseur de réseau vectoriel (VNA).

Plasma activated mist: Characterization and application efficacy

As the demand for healthier and ‘clean label’ food products is ever increasing, there is a need to develop novel food processing methods to address the risks associated with food safety while ensuring sustainability. Cold plasma activated water (PAW) and plasma activated mist (PAM) are gaining attention for potential applications in food products as a ‘green disinfectant’. The intern will conduct research with a dielectric barrier PAM generation system where the process conditions will be optimized for maximum reactive species concentration. Further, the antimicrobial properties of PAM will be evaluated with common food pathogens such as Listeria monocytogenes, Salmonella and Escherichia coli. The effect of the plasma activated mist treatment on the quality attributes of food materials will also be studied. The research findings generated by the intern will help the partner organization to scale-up and commercialize their plasma activated mist technology.

Methods for Paramagnetic and Diamagnetic Susceptibility Separation

Recent advances in Magnetic Resonance Imaging (MRI) have enabled visualization of iron and myelin in the brain using a method that can measure the effects of MRI field changes. The physical property measured is called magnetic susceptibility, which can visualize iron accumulation in the brain and also myelin which is a wrapping around nerves. The partner organization has implemented software to detect this effect. However, this method cannot separate out the individual contributions of myelin and iron. In diseases such as Multiple Sclerosis, iron and myelin changes can happen at the same time, thus separating them out is very important. The research project will involve an expert in magnetic susceptibility who will test methods for susceptibility separation and develop best techniques for performing this task, then implement them on the partner’s MRI systems. The benefit to the partner will be new and proven reconstruction techniques to further advance MRI.

Projet rétention et valorisation des étudiants Internationaux

L’UQTR, cherchant à attirer davantage d’étudiants internationaux, évolue dans un environnement concurrentiel et veut se démarquer auprès des étudiants internationaux. À cette fin, en partenariat avec IDÉ Trois-Rivières, les deux établissements épauleront une personne stagiaire ayant pour mandat le recensement des outils et ressources actuelles. À la fin du projet, la personne stagiaire proposera une stratégie afin de favoriser l’attractivité et l’intégration, puis la rétention des étudiants internationaux dans la ville de Trois-Rivières.

Hover BSI Internshpi

Hover seeks to recruit an intern to aid in the evaluation and enhancement of Drone Delivery within Canada. This internship plays a vital role in fostering the growth of the industry within the country. Should the minimum viable product (MVP) demonstrate a favorable response, it will pave the way for the advancement of this novel technology across various Canadian enterprises. The intern will contribute to the establishment of partnerships with businesses intrigued by this technology, as well as assist in drone operations. By receiving assistance and insights, Hover stands to gain substantial benefits. Given our status as a small company, the intern will have the opportunity to actively participate in decision-making processes that directly shape the company’s future.

Learning to organize and discover biomedical scientific literature

This project will investigate machine learning methods to organize and discover the vast literature on biomedical science. First, it will focus on named entity recognition–the task of finding and classifying entities in text documents–on large collections of abstracts and full-text of research papers and investigate semi-supervised learning methods to leverage large collections of unlabeled research papers. Second, it will focus on designing new ranking algorithms for research papers by exploiting information from several sources, including citation networks of papers, future impact predictions of new research papers and domain ontologies. The solutions to these problems will directly impact the performance of the methods that are currently in use by the company to solve the discovery problem of biomedical research papers.

Feasibility of Hybrid Ground-Source Heat Pumps for Sustainable Heating in Cold Climates

Using fossil fuels in building heating systems is greenhouse gas (GHG) intensive, leading to global warming. Ground-source heat pumps (GSHPs) are cleaner electric alternatives to fossil fuel systems. They utilize and
transport heat from the ground to warm buildings. However, prolonged use of GSHPs can lead to a phenomenon known as thermal imbalance, which depletes the ground heat content and decreases soil temperature reducing
the performance of GSHPs. Additionally, GSHPs are expensive compared to traditional fossil fuel-based heating systems. To address these issues, this study explores the feasibility of a hybrid GSHP system, which combines a
GSHP with a natural gas furnace. The aim is to optimize the operation of the hybrid system to have minimal initial/operating cost, and emissions while maintaining high energy efficiency.

Entraînement d’un agent conversationnel pour les procédures internes

Koïos Intelligence est à la source de l’agent conversationnel Olivo qui vise à offrir une expérience
interactive, guidant l’utilisateur au travers des processus de prévente, de vente et d’après-vente pour
tous les types d’assurances. Bien que l’outil soit dans un état avancé tant au niveau de la conversation
écrite qu’orale, et ce aussi bien en français qu’en anglais, son amélioration se heurte aux exigences
computationnelles lourdes pour l’entraînement des modèles d’apprentissage automatique sous-jacent.
De plus, les outils de discussions interactifs sont souvent imprécis dans leur développement et mise en
oeuvre dans le cadre d’un champ d’application particulier. Dans le domaine de l’assurance, il y a peu de
solutions adaptées aux procédures internes. L’idée serait donc de perfectionner un robot interagissant
avec les employés et, selon leur position, les accompagner dans certaines tâches administratives
(formation d’un nouvel employé, accès à la documentation interne, etc.). Il est nécessaire pour les
rendre plus pertinents d’améliorer leur entraînement en utilisant des bases lexicales spécialisées, dont
dispose en interne Koïos Intelligence.

Identification et suivi de la performance des athlètes d’excellence du baseball québécois âgés de 15-18 ans

Une étude transversale et une étude longitudinale seront réalisées pour créer des profils de performance chez les meilleurs athlètes de baseball de la province. Ces profils basés sur des données anthropométriques, physiques et psycho-physiques permettront de distinguer les athlètes les plus talentueux et d’aider à leur sélection dans le programme d’excellence de l’Académie de Baseball du Canada. De plus, ces informations contribueront à améliorer les évaluations, les programmes d’entraînement et le développement des jeunes athlètes.
Le partenariat entre l’UQTR et l’Académie de Baseball du Canada (ABC) vise donc à améliorer l’encadrement des athlètes et à évaluer leur performance à court et à long terme. Une collecte approfondie de données et la création d’une base de données seront essentielles pour ce suivi des athlètes de haut niveau au Québec.