Innovative Projects Realized

Analyzing noise compensation properties of trained recurrent neural networks

Reliability is a fundamental requirement for computational systems, brains and artificial models alike: a system should respond the same way for repeated presentations of the same stimulus. However, the brain has two features that can threaten its reliability: intrinsic stochasticity and chaos. Stochasticity takes the form of random fluctuations affecting the reliability of components of the system, whereas chaos is an emergent property of the entire system that causes similar inputs or initial conditions to produce totally different outputs. The brain must have mechanisms to compensate for its noisy and unreliable machinery, and the goal of our project is to characterize these mechanisms. To this end, we will first develop tools for quantifying the reliability of models of neural circuits, and we will subsequently extend the capabilities of these tools to analyze data collected from real neuroscience experiments.

Translation for educational change: Translator identity and the Global Storybooks project

Our research seeks to better understand challenges that translators encounter in translating stories for the diverse sites on the Global Storybooks portal (https://globalstorybooks.net/), and how they resolve these challenges. We will draw on our existing database of translation data to investigate how translators draw on a range of linguistic resources to achieve a satisfactory representation of the target translation. Practically, this data will help us to develop improved guidelines for translators, as well as proofreaders, editors, and coordinators. More theoretically, the data will enable us to develop a model of translator identity that will legitimize the diverse participants who are active in translation projects. We hope this model will provide a useful framework for further research on translation in diverse global contexts. The sub-objectives of the project are as follows: Sub-objective 1: Review existing database of translation data in the Global Storybooks project and identify common translation challenges and practices of translators. Sub-objective 2: Identify two translation case studies for comprehensive analysis. Sub-objective 3: Based on the case studies, (i) develop a comprehensive set of guidelines for translators working on new Global Storybooks sites and (ii) develop a model of translator identity that would be useful for further research purposes.

Decoding the neural dynamics of emotion-related human memory optimization using AI-informed multivariate techniques

Episodic memory, our fascinating ability to encode and mentally relive past experiences, lies at the core of human cognition. It allows individuals not only to recall past events, but it is crucial in planning and guiding future behavior. However, among all of our daily-life experiences, only some events will be transformed into lasting memories, particularly because of their emotional salience. From the brain perspective’s, emotions are thought to sustainably recruit the amygdala, thus facilitating memory processes occurring in the hippocampus. In addition, while the first step in forming durable memories is the initial learning, it has become increasingly clear that memory reinstatement of encoding neural patterns at retrieval also plays an important role in remembering. Yet, we know very little about the precise brain mechanisms supporting emotion-related memory enhancement via neural pattern reinstatement. Here, we propose to use machine learning to establish an AI-informed multivariate approach to isolate the neural processes that are boosted by emotions and which mediate learning and retrieval enhancement.

Decoding the neural correlates of dynamic decision-making in humans

This research project will combine computational modeling, machine learning (ML) algorithms and whole-brain neural recordings (magnetoencephalography, MEG) to shed light on how the mechanisms underlying dynamic decision making are implemented in the human brain. Specifically, we will use the statistical framework of information theory to characterize inter-areal neural coupling and the direction of information flow when weighing sensory evidence and committing to a specific choice during dynamic decision-making. By providing a quantitative link between the behavioral and neural dynamics subserving how decisions are continuously formed in the brain, this project will contribute to expose mechanisms that are likely to figure prominently in human cognition, in health and disease.

Sliammon Carbon Interests: Background research on feasibility of acarbon offset project application.

This research project will explore the costs and benefits of putting a carbon offset project into

application for Tla’ Amin Timber Products. This project will prepare the background and inventory

documents for a carbon offset application according to the two chosen Standards, Protocol for tbe

Creation of Forest Carbon Offsets in British Columbia Version 1.0 for ImproVed Forest

Management and Pacific Carbon Standard. Simultaneously, I will develop a community

engagement program so that I can provide basic carbon offset and forest management terminology and

knowledge. I hope to provide enough information throughout the life of the project so that

stakeholders understand the project findings; and stakeholders are able to make a fair decision about

project acceptance or refusal. Up to 4 carbon offset project scenarios will be presented to Tla’Amin

stakeholders at a community presentation at the end of 6 months. I will use a number of softwares and

programs…

Biosensing JFET platform with printed graphene gate and customizable functionalization

The goal of this research project is to create a novel type of biosensor by combining two complimentary microfabrication techniques. First, a silicon chip containing JFET transistors with an open gate will be fabricated using traditional microfabrication techniques that are highly reliable and give good performance. Second, a graphene layer will be inkjet printed onto the open gate of the transistor. The graphene will act as a sensor transducer to sense pH or biological species such as insulin. Graphene has a large surface area and good electrical properties, which makes it the ideal transducer material. The graphene can be functionalized to detect various different biological species. Combined with the flexibility of inkjet printing, this will allow different biosensors to be fabricated easily next to each other on the same silicon platform.

Définir les objectifs et indicateurs d’adaptation aux changements climatiques

Puisque les changements climatiques sont déjà ressentis et le seront de plus en plus, il est nécessaire de s’adapter à de conditions environnementales sans précédent. Une quantité grandissante de programmes de financement vise maintenant à soutenir les communautés dans leurs démarches d’adaptation; par conséquent, le besoin d’évaluer l’atteinte des cibles d’adaptation devient de plus en plus important. Étant donné que le concept de l’adaptation est en évolution constante, et que les mesures d’adaptation sont généralement plus difficiles à quantifier que celles de lutte (par ex., réduction des émissions carboniques), il est essentiel de définir des indicateur quantifiables pouvant être utilisés par différents paliers de gouvernements (municipalités, régions, etc.) pour mesurer l’atteinte des cibles d’adaptation. L’objectif principal du présent projet est donc de recenser les cibles et les indicateurs d’adaptation aux changements climatiques utilisés à l’échelle mondiale afin de déterminer lesquels pourraient être applicables au contexte québécois. Ce projet aura permettra d’avancer la recherche dans ce domaine et d’interagir avec les acteurs du milieu, et donc permettra de faire rayonner l’organisme partenaire en ce sens.

Analyse et recherche d’optimisation d’un processus d’adaptation de contenu issu de la numérisation 3D à un contenu photo-réaliste et interactif

Le récents progrès dans le domaine de la numérisation 3D (SLAM, Lidar, lumière structurée, photogrammétrie, etc.) ont ouvert de nouvelles portes quant à la possibilité de réaliser des maquettes numériques interactive. Cependant, les outils traditionnels pour traiter les données brutes (nuage de point) de ces capteurs ne permettent pas de générer un modèle numérique adapté à l’utilisation d’une maquette 3D numérique photoréaliste et interactive. PreVu3D désire donc mettre en place son propre processus de traitement de donnée en adaptant les outils actuels et, dans le cas où ces outils n’existent pas, développer ces propres outils adaptés à ce cas particulier.

Greening the Canadian Landscape

Urbanization is rapidly increasing in Canada and these areas are confronting major challenges associated with climatic change. Canadian cities must navigate these challenges and may also significantly contribute to mitigation efforts. However, municipalities and developers lack the state-of-the-art information, policies, and resources necessary to be successful in this endeavour. This research project uses a case study approach to establish state of the art in relation to four themes (low-impact development, stewardship, community well-being, natural assets), consider each in terms of resilience, establish an evidence base of best-practice, execute active experimental trials, and monitor/evaluate their performance. The case study focuses on Prudhommes Landing, a significant Lake Ontario waterfront development. Knowledge generated will be transferable to developments elsewhere, provide a basis for evidence-based decisions, and influence the uptake of best practices. The project will contribute to Vineland’s core mandate for applied research and enhance their industry leader position in resilient landscapes.

Development of LBT-1 as an alternative technology to antibiotic molecules

Antibiotic resistant bacteria are one of the major threats of the 21st century. These superbugs, which are resistant to virtually all our antibiotic arsenal, turns benign infections, such as urinary tract infections, into life threatening diseases. It is estimated that, by 2050, 10 million people will die each year because of antibiotic resistance. We thus need to develop alternative technologies to conventional antibiotics in order to prevent this scenario from unfolding.
This project proposes a totally new approach to eliminate antibiotic resistant bacteria. Instead of using molecules to kill bacteria, we propose to use engineered probiotics capable of recognizing and eliminating specifically antibiotic resistant bacteria. This therapeutic breakthrough approach turns superbugs into regular bugs that we can eliminate.

Assessing the Sustainability of Snowmobiling Trails across Ontario

The proposed research seeks to understand how marginal winter conditions have impacted Ontario snowmobile trails in the past in order to make inferences about the future. The project will combine historic climate data (1989-2019), operational indicators (trail opening/closing, permit sales), and Regional Climate Model (RCM) outputs for the Province of Ontario to explore the implications of persistent marginal snowmobiling conditions and/or loss of district trails on long-term participation in the sport. The results of this study will directly benefit the partner organization by providing insight into which district trails will be climatically viable in the future, along with insight into shifting snowmobiler demand patterns under warmer conditions. The results will allow industry managers to prepare for supply- and demand-side changes in both the near-term (2050s) and longer-term (2080s), guiding decision-making on how best to invest and distribute limited resources (e.g., groomers, fuel, infrastructure) across Ontario’s 30,000km trail network.

Research into Convolutional Neural Network (CNN) Explainability

Machine Learning is advancing at an astounding rate. It is powered by complex models such as deep neural networks (DNNs). These models have a wide range of real-world applications, in fields like Computer Vision, Natural Language Processing, Information Retrieval and others. But Machine Learning is not without some serious limitations and drawbacks. The most serious one is the lack of transparency in their inferences, which works against relying completely in these models and leaves users with little understanding of how particular decisions are made. In this research we will explore new ways to represent the evolution of a CNN during training, and how the artifacts generated by these representations can be traced back to the inputs they’ve used during training. We will also explore the issue of imbalance in datasets, from the perspective of Parallel Coordinates and how to visualize dataset imbalance using this technique. The preliminary prototypes are going to be measured through usability tests with experts within the partner organization.