Projets novateurs réalisés

Efficient Learning Methods for Multimodal Understanding

The main focus of this research is to develop representation learning architectures and algorithms that can help perform various multimodal understanding tasks, and at the same time reduce the need for human supervision in the form of costly annotations. To achieve this goal, a learning system must be able to: (1) learn new tasks or concepts with a few examples; (2) effectively use the knowledge already acquired by the system; (3) rely on one modality (e.g., text/audio) to fill in gaps in another modality (e.g., vision); and finally (4) retain high performance on all tasks/concepts learned previously. The goal of this research is to develop methods that enable efficient learning for multimodal tasks (e.g., dialogue-based image/video retrieval) in scenarios with limited annotated data. As part of this work, we will focus on answering two key questions: (1) How to leverage the problem structure in order to enable learning algorithms that use limited annotations more effectively? and (2) What are the mechanisms that enable efficient learning from few examples?

Advanced Text Clustering Algorithm for Aircraft Applications

CaseBank provides a service called ChronicX™ to the airline industry for the purpose of detecting and managing repeat defects, i.e. faults that have eluded resolution repeatedly. Each night, airlines upload their raw maintenance records to the CaseBank server. ChronicX applies text mining methods to eliminate irrelevant records, and search for defects that are repeat occurrences of defects previously reported. These are assembled into clusters, each of which is called a ‘chronic’. ChronicX performs reasonably well, but it has limitations that we believe can be improved.

Marketing

The pandemic negatively impacts most businesses and this will likely influence consumer behavior in the future. Habitat for Humanity aims to adapt to the new situation and increase their revenue by focusing on some business strategies. These strategy require comprehensive research in terms of the real estate industry, market environment, market share, etc. It is projected by using initiative strategies and targeting specific institutions, HFHMVI as a non-profit organization will be able to fund its projects and shelter more people.

La gestion du climat

La mission de Muclitech est de concevoir, produire et fournir des équipements de gestion du climat se démarquant par leur ingéniosité, leur adaptabilité et leur qualité, répondant en cela aux besoins et attentes d’une clientèle nationale et internationale à la recherche de solutions innovantes.

Principales activités de Muclitech
Accompagner le producteur agricole dans la conception de ses infrastructures de production en milieu clos.
Fournir et installer de l’équipement pour une salle de culture en milieu clos.
Accompagner le producteur dans la gestion de l’exploitation agricole sur le long terme.

Grâce à la participation de la stagiaire, Muclitech espère développer une opportunité d’affaires qui se développe au Québec et qui a pris encore plus d’ampleur depuis la pandémie de Covid-19. En effet, le secteur de la production de légumes et de fruits en milieu clos est présentement en expansion. La pandémie a créé un mouvement favorable pour l’achat local de ces denrées. Muclitech espère se tailler une place de choix dans ce secteur.

Développement de vaccins vivants atténués contre les infections causées par les Escherichia coli pathogènes aviaires

L’utilisation abusive et la mauvaise utilisation des antibiotiques chez les animaux et les humains contribuent à la menace croissante de l’émergence des souches résistantes aux antibiotiques. Certains types de bactéries qui causent des infections graves chez l’homme et l’animal y compris les poulets, les bovins et les porcs ont déjà développé une résistance à la plupart ou à tous les traitements disponibles, et il y a très peu d’options prometteuses. Le développement des approches alternatives à l’utilisation des antibiotiques en industrie avicole reste une priorité majeure afin de mettre fin à l’utilisation des antibiotiques. La vaccination des volailles à l’aide de vaccins vivants atténués contre la colibacillose devient de plus en plus nécessaire pour maintenir le bien-être des animaux et réduire la quantité et la fréquence d’utilisation d’antimicrobiens dans les fermes.

Middleware-Layer for Mobile Game Architecture

Mobile devices are well established at the heart of our digital lives. Today, these devices are utilized as personalized and enterprise application nodes. Moreover, mobile devices support ubiquitous access to data which even makes them a pivotal component in the deployment of applications such as online games. However, there are challenges that have plagued the mobile field. No matter the advances we make, mobile devices still rely on wireless channels to communicate, and these channels experience sporadic disconnections and bandwidth fluctuations. Hence, activities such as online gamification using the mobile as the consumer node experiences setbacks such as loss of data and poor data transmission. In this research, we aim to enhance the mobile gamer experience by proposing a middleware framework that facilitates real time communication exchanges including support for high user throughput.

Use of Nanocellulose for Treating Fines Dominated Oil Sand Tailings Deposits

Reclamation of oil sands fluid fine tailings (FFT) ponds is challenging due to the slow rate of natural consolidation of the suspended sediments generated during the extraction process. In order to reclaim tailings ponds, the shear strength of the sediments must first be increased to make the surface trafficable, at which point further reclamation work can be carried out on the surface of the ponds. This work presents an exciting opportunity to investigate the incorporation of nanofibrillated cellulose (NFC) into oil sands FFT to improve the strength of the tailings and expedite the remediation timeline. Successful demonstration of the practical application of NFC for tailings strengthening would result in a new market for Canada’s forestry sector, while simultaneously helping to address the ongoing tailings remediation and reclamation challenges experienced by Alberta’s oil sands industry.

Obfuscation and Homomorphic Encryption in the Quantum Setting

Our research goals, which are a mix of theoretical and practical problems, are motivated by the practical near-term problem of delegating tasks to an untrusted remote quantum computer in a secure way. Toward this, we consider the notions of indistinguishability obfuscation for quantum circuits and quantum fully homomorphic encryption schemes. The former refers to algorithms that take a given quantum circuit and make it less intelligible, such that the new circuit still gives the same outputs as the original one. The latter refers to encryption schemes that allow a remote party to perform any arbitrary computation directly on encrypted data, without being able to decrypt it. Our goal in the project is to advance the current theoretical understanding of the above-mentioned notions and apply our results to the “secure delegation of tasks problem”, enriching the partner organization’s knowledge and products.

Online Public Consultation in the Resource Development Industry

Over the years, the methodology for public consultation has evolved to include public information meetings and hearings, telephone polls and surveys, and most recently online engagement. Increasingly, government is emphasizing evidencebased decisionmaking. They recognize that existing methods are lacking in reaching a broad demographic. Internet communication technology has made it possible to reach a broader
spectrum of the public however until recently, that interaction has been anonymous and anecdotal at best. The objective of the proposed research is to explore the use of online public consultation technology to advance best practices within the real estate development industry. The project aims to investigate the ways online consultation generally, and the PlaceSpeak platform with locationbased authentication in particular, can enrich public development consultation bringing together current and traditional outreach methods.

StreamSight: Deep Learning Techniques for Managing Contaminants in Residential Curbside Recycling

This project works to classify contaminants found in residential curb-side recycling. This is done automatically using computer vision techniques. As recycling is tipped into the recycling truck, cameras take pictures of the recycling and computer vision software works to identify contaminants found in the load. With municipalities equipped with this fine-grained data, they will have the ability to produce targeted education campaigns to improve the recycling process and reduce contamination found in recycling. Image data collected from household recycling will be stored securely and images are not captured with the household’s address or any personally identifiable information connected with the image.

Process study, simulation and optimization of direct conversion of biomass to ethyl evulinate

Increasing concerns about global warming related to greenhouse emission, and depletion of fossil fuel resources, bring biomass as a promising environmentally friendly and sustainable alternative to supply chemicals and fuels. One attractive option is the conversion of biomass obtained from recycling activities to high-value chemicals. In this context, direct biomass conversion to ethyl levulinate (EL) was successfully demonstrated [1] in a small batch reactor through a one-step acid-catalyzed biomass hydrolysis-esterification process to EL using 1, 2-dichloroethane (DCE), in the objective of obtaining an economical process. EL has prevalent applications in the fragrance and the flavour industries and is used as an additive for diesel and biodiesel [2,3,4].
A probable and proposed reaction pathway for the reaction comprises three subsequent steps including 1) hydrolysis of glucose to 5-chloromethyl furfural (CMF), then 2) hydration of CMF to levulinic acid (LA) and formic acid (FA) and finally 3) esterification of LA and FA to EL and ethyl formate (EF) [1, 3, and 4].

ResolvMD – Business Development Analyst

ResolvMD is a medical software startup that is focused on building a disruptive solution to a multi-million dollar problem. Our product is a best in class billing solution that will serve to re-define the value proposition for how physicians understand their practice in a thoughtful, insightful and transparent way. As we transition technology along the spectrum from analog to digital, we need to understand each potential client in order to make an intuitive interface that facilitates a smooth, efficient and personalized experience. An intern will be instrumental in helping us to understand the optimal UI / UX by engaging with the target market constituents to glean a thorough understanding of their needs and then working with our tech / management team to translate that knowledge to our software. The benefit to our organization from doing this properly is an ability to deliver an experience that is absent from the market today – solidifying us as leaders and innovators in the space.