Multi-source data fusion to assess the impact of roadworks on secondary travel demand

Since recent years, many roadwork projects and nearby events will affect the flow of visitors at Casino de Montreal, a model is needed to analyze the impact of these roadwork events and provide solutions to mitigate the effect. In this project, multisource data will be provided and applied into traffic flow prediction. The main objective of this project is to build a predictive model and then using the results of this model to perform analysis.

Integration of hippocampal subfield measurements into a streamlined MRI volumetry pipeline using deep learning

The hippocampus is a small brain region that has a key role in brain diseases including Alzheimer’s and epilepsy. The measurement of hippocampal volume on MRI is important for the diagnosis of these diseases, but this is a complex task because it is a small structure. Further, the hippocampus is divided in sub-regions that are impacted differently by the diseases. This project aims to use artificial intelligence technology to rapidly and reliably measure all sub-regions of the hippocampus and develop this as a clinical tool that could be used in hospitals and clinics.

Mesurer l’impact socio-économique d’initiatives en éducation financière dans un pays en voie de développement : Méthodologies proposées

De nombreuses ONGs implantent des plans éducatifs dans des pays en voie de développement. Habituellement, des indicateurs de l’atteinte des objectifs sont proposés afin d’évaluer les retombées directes du projet. Mais qu’en est-il des retombées indirectes auprès des populations impliquées dans le projet? Comment utilisent-elles les nouveaux apprentissages dans les autres domaines de leur vie? Comment évaluer ces retombées indirectes lorsque des participants adultes ne savent pas lire ni écrire?

Covalent and non-covalent interactions self-supervised representation of molecules for chemotherapeutic drug design

Most of the drugs used to treat cancer have been originally identified from natural sources. While Nature did a great job selecting those compounds, some of them have shown limitations in the treatment of cancer and others have shown to be insufficient on some cancer types. Furthermore, it exists a gigantic number (nearly infinite) of small molecules human can synthetize.

Milk spectral signature of transition success or failure

A successful transition to lactation is of high importance in dairy production as this period affects the entire productive life of a dairy cow. Poor transition are often associated with clinical or subclinical diseases. The objective is to assess whether milk components are associated with a poor or good cow transition. Emphasis is put on milk components other than the ones routinely measured in monthly milk samples. The spectral signatures of milk from cows analyzed through infrared spectroscopy are therefore assembled.

Multi-scale direct measurements and source determination of methane emissions in Montréal

Methane is a potent greenhouse gas and reducing methane emissions is a key strategy in many climate change action plans. Because cities and their methane emissions are growing, it is essential to characterize and mitigate methane emissions from urban sources. Because there are no published studies of methane emission measurements in Montréal, we propose to conduct direct measurements to quantify methane emission rates and attribute the emissions to sources in Montréal.

Separating Syntax and Semantics for Semantic Parsing

Study of language disorders, theoretical linguistics, and neuroscience suggests language competence involves two interacting systems, typically dubbed syntax and semantics. However few state-of-the-art deep-learning approaches for natural language processing explicitly model two different systems of representation. While achieving impressive performance on linguistic tasks, they commonly fail to generalize systematically. For example, unlike humans, learning a new verb like jump in isolation is insufficient for models to combine it with known words (like jump twice or jump and run).

Humic Land, a biological promoter of crop growth and the soil microbiome

Humic Land is a multi-purpose, 100% organic fertilizer that was produced from black peat using innovative technology that protects live soil microorganisms. It contains a microbial consortia that may produce plant-growth promoting substances, thereby acting as biological promotor of crops growing in stressful conditions.

Few-shot Generative Adversarial Networks

The most successful computer vision approaches are based on deep learning architectures, which typically require a large amount of labeled data. This can be impractical or expensive to acquire. Therefore, few-shot learning techniques were proposed to learn new concepts with just one or few annotated examples. However, unsupervised methods such as generative adversarial networks (GANs) still require a huge amount of data to be trained. As such, this project will focus on few-shot learning for GANs.

Effect of chemical composition on machining of Inconel 625/718 components by machine learning and microstructural analysis

APN in Quebec City faces a high fluctuation of the lifespan of tools in machining operations such as milling and turning. This hampers production planning, since the number of necessary tools for a job as well as their lifetime cannot be well predicted. Previous investigation on this challenge could not determine a reason for this phenomenon. Therefore, the objective of this project is to identify a correlation between the available process data such as material properties and machining parameters with the length of tool life.