Projets novateurs réalisés

Interpretability and explainability of Deep Neural Networks

In recent years ·complex deep neural networks have achieved outstanding performance on challenging tasks such as language models, classifiers and machine translation but they have remained as a mystery for the users. Due to their hidden and difficult to comprehend internal structure, as well as to their sheer size, they are often referred to as “black box” models. Additionally, the widespread adoption of machine learning algorithms has increased the necessity to trust to these moging dels in order to employ them for decision-making in critical situations. At the same time, making critical decisions concerning humans without understanding the justification of such a decision is unacceptable, both ethically and legally. Therefore, there is an increasing interest in the Machine Learning community in interpreting or deriving explanations able to describe a black box’s behavior. The need to understand black-box models’ decisions has resulted in the growth of research on their explainability and interpretability. For this research we are planning to design mathematical methods to explain the behavior of these models .

Quantifying the relationships between rail profile quality indices and inservice wheel-rail contact conditions

In order to maintain safe and efficient operations, railway systems periodically use large, special purpose, trackbound grinding machines to re-establish the desired profiles of rails, and to remove surface damage that has occurred under operating conditions. The accuracy of the grinding process is typically measured using quality indices to compare the resulting rail profile to its target shape. In this research project, the intern will develop quantitative methodologies (incorporating real-world grinding results and measurements, together with detailed modeling and simulation tools) to better understand the relationships between the quality indices that are calculated, and the actual wheel-rail contact conditions that they are intended to represent. This will provide the basis for economic modeling that helps the industrial partner (ARM) work together with its customer base to make informed decisions when establishing maintenance targets and planning maintenance programs.

Apprentissage automatique pour la résolution de problèmes de repositionnement de véhicules

L’entreprise Fastercom se spécialise dans le développement d’algorithmes et de logiciels permettant de résoudre différentes formes de problèmes liés aux transports. L’objectif de ce projet est de développer des techniques d’apprentissage automatique adaptées au problème de repositionnement dynamique de véhicules. Il s’agit par exemple du repositionnement d’ambulances pour couvrir adéquatement une zone pour les urgences, ou repositionner des taxis pour pouvoir accueillir de nouveaux passagers. L’objectif est de répondre à des demandes de course en temps réel et d’attribuer les véhicules disponibles aux courses sollicitées. La gestion de la flotte demande de sélectionner un véhicule pour une course et de repositionner les véhicules efficacement en attente des prochaines courses.

Application du laser terrestre et modelisation 3D en archeologie

Le present projet vise a democratiser l’utilisation du laser terrestre et de la modelisation 3D en archeologie. Souvent consideree com me couteux ces outils sont tres peu souvent envisage dans un contexte de recherche archeologique. Or, depuis une decennie la revolution numerique que nous connaissons a enormement simplifier I ‘utilisation du laser terrestre en plus d’ augmenter ses capacites. De plus, Ies logiciels de modelisation 3D sont desormais beaucoup plus ludiques et nettement plus abordables. Pourtant, malgre ces avances leur utilisation demeurent tres tim ide en Amerique du Nord. Nous souhaitons done par ce partenariat presenter les avantages que l’archeologie retirerait d’une utilisation plus systematique du laser terrestre et des modelisations 3D.

Fleet Management Strategies

The project will involve analyzing the fleet and exploring the costing, Reporting on inventory and assets optimization strategies using industrial accepted and academic approaches. The intern would factor in real time pressures and document those to allow for a Strengths -Weakness- Threat and Opportunity Analysis.
The conclusion of the research will allow the company to better plan their operations and look for growth in the region

Method Development for Systematic Analysis of Anaerobic Bacteria in Marine Water/Soil Sediment at Atlantic Offshore

A chemical process that causes decay in materials is known as Corrosion. The rate and degree of corrosion depend on the immediate environment affecting the materials. Offshore oil and gas industries are equipped with subsea pipelines, cables, vessels, etc. The presence of microbes in the production stream or the transport line may cause corrosion in the subsea assets and other marine equipment. Corrosion triggered by the presence of microbes is called microbiologically influenced corrosion (MIC). In this project, the intern will develop and test analytical methodologies to assess the degree of microbial activity in different marine samples (water/soil sediment). This will help to assess the chances of MIC on an asset’s surface in a given environment.

qualiTEAS, is a corrosion inspection and management company based in St. John’s, Newfoundland. We have a proactive approach, unlike other service companies, where we investigate the mechanism of corrosion by identifying different environmental factors that are responsible for causing corrosion. Thus, through this project, developed analytical methodologies for microbial analysis can assist to understand the impact of microbial activity in the corrosion process. This will further complement company’s service scope for the corrosion inspection of subsea assets.

The use of thermographic imaging in the detection, identification and characterization of Deep Venous Thrombosis (DVT)

The project will investigate the relationship between Deep Vein Thrombosis (DVT) and temperature. The intern will use researched knowledge and thermal metrics to design a simulated trial to gather data (acquiring and processing several images), analyze (reviewing data and performing analytics), model (create specific DVT data models) and validate the simulated data with real data (human trial phase) for the thermal imaging of DVT. The intern will carry out this experiment using a novel 3d thermal imaginer. Results from this study will be used to prepare for the human participant trial; thermographic imaging detection of DVT in the lower extremities. Outcomes from this project could provide an application to partake in the Venous Thromboembolism treatment market valued at over $950 million in 2020 and forecasted to be over $1.7 billion by 20

Quantifying spatio-temporal variability of fire fuels in post-mountain pine beetle outbreak in Jasper National Park using terrestrial and airborne laser scanning and RPAS structure from motion

Well-known Canadian montane national parks have undergone a length period of fire suppression activities, which
started in the early 20th century. This has resulted in old growth forest stands that are relatively homogeneous in
nature, often containing forests that have the same species, age, and structure. In recent years, mountain pine
beetle outbreaks have occurred, especially in Jasper National Park, resulting in the dramatic transformation of
lodgepole pine forests surrounding the town site. The mortality of high proportions of trees attacked by mountain
pine beetle have resulted in a shifting of dead materials as trees eventually break down and fall over, increasing
the risk from wildfire as dry, dead forest fuels accumulate. This study uses novel light detection and ranging
(LiDAR) technologies and remote piloted aircraft systems (RPAS) to collect highly detailed, three-dimensional
structural information of dead and dying forest structures to determine how fuels for wildfire vary over space and
through time as forests are attacked, die, and eventually break down.

Finite Element Assessment of Novel High-Performance Building Panels

Rising material costs, lack of quality control, labour shortages, challenging climates and significant on-site waste continue to be the challenges encountered by the Canadian construction industry. The collaboration between the research team and NEXII will develop the technical know-how for the new generation building panel products to address these challenges. The new NEXII panel is a three-layer composite “sandwich” structure consisting of an expanded polystyrene insulation (EPS) core between two layers of Nexiite, an innovative material produced by NEXII. The application of the new generation building panel products leads to: (1) cleaner construction sites; (2) less noise and neighborhood disruption; (3) fire, water and mold resistant buildings; and (4) healthier indoor spaces. Preliminary structural assessment has already confirmed the feasibility of the research project. Advanced computer simulation using the finite element (FE) technique will be performed to optimize the structural and building envelop performance of the new products.

Investigation on the Effects of Water-in-Oil Emulsion on CSI Recovery Performance from Microscopic Perspectives

This research work will establish a systematic workflow for analyzing transient equilibrium foamy oil phase behavior by coupling the m-CCEC tests, depletion rate and presence of water-in-oil emulsion from microscopic viewpoint which are seldom performed for heavy oil. It will provide a strong connection and comparison with previously studies which was conducted in the absence of water-in-oil emulsion and in PVT system. It will create a strong connection between phase behavior with fluid properties, operating conditions and kinetics. A concrete database which contains sufficient amount of laboratory transient microflow data for selected heavy oil reservoir can be established.

Warr’s Pharmacy Scope of Practice

This project looks at innovative ways to expand the services the industry partner offers and to innovate for the needs of the community. It will be it approached looking at the best way to expand the scope of services they offer that conform with pharmaceutical standards and industry regulations.

Investigating the feasibility of and theimplementation of preimplantation genetictesting at Atlantic Assisted ReproductiveTherapies

Atlantic Assisted Reproduction Therapies (AART) is the primary facility located in the Atlantic Provinces capable of assisting reproductively challenged couples. Fertility Pharmacy Services(FPS) is an onsite subsidiary of AART. AART offers a wide range of services, however, AART currently does not have the capacity to assist individuals who require preimplantation genetic testing. Clients who require such genetic testing include persons with known genetic conditions and/or individuals who have suffered repeated miscarriages. The overall objective of this project is to investigate the feasibility of establishing a preimplantation embryonic genetic screening program at the AART center, located in Halifax. The project will be carried out through a number of stages, from researching the genetic tests best suited to FPS/AART’s clients, to establishing the equipment and facility, to finally performing new services for FPS/AART clients. This project will be highly beneficial to the region as it will allow AART to assist a new group of clients, therefore building the clientele serviced by the FPS