Modern machine learning is powered by deep neural networks composed of many interconnected layers of artificial neurons, whose tunable connections learn from data to solve important problems. While this approach has achieved incredible successes in many domains, in practice neural networks act as "black boxes" whose high-level insights are hard to access. Our project will study the use of a promising new family of tensor network models, originally developed for learning high-level structure in complex quantum systems, for capturing the structure of natural languages like French or English.

This project is addressing the problem of portfolio optimization by using reinforcement learning, an area of machine learning that has recently attracted many researchers. Its advantages compared to the conventional models of portfolio optimization are coming from its ability in incorporating many features of the assets into the asset allocation problem without relying on the predicted returns outputted from another model. This model is able of considering risk measures and providing asset allocation strategies that are in line with the risk preferences of the investors.

The proposed research will enable customers to see personalized recommendations based on multiple factors such as their order history, their preferences and contextual information such as the meteo and the day of the week. The main expected benefit is to increase the average bill by showing personalized recommendation

Le FerroSilicium (FeSi) est produit dans un four à arc à l’aide de plusieurs matières premières qui sont le quartz, le charbon, le bois et la ferraille d’acier. C’est un procédé complexe à opérer puisque les matières premières sont réparties inégalement dans le four, il est difficile de mesurer des données à l’intérieur du four à 2000 degrés Celsius et le four est présentement opéré selon l’expérience des ingénieurs et des opérateurs. Le but de ce projet est de créer un four numérique, c’est-à-dire un logiciel qui simulera une fournée de FeSi.

La rotation de personnel est un problème majeur dans l’industrie des soins de santé à domicile. Dans un contexte où le travail des intervenants est particulièrement basé sur une relation de confiance avec les patients, la prévention de la rotation est déterminante pour le continuum et la qualité des soins prodigués.

Electricity is a commodity traded in dedicated markets. Electricity producers, distributors, brokers, speculators and industrial consumers participate in these markets to match supply with demand, hedge their exposure to electricity prices and/or profit from dislocations in the price between geographies or time horizons. Accurate price forecasts enable market participants to make better business decisions. BCM Energy is a vertically-integrated electricity distributor and broker in the European markets.

The project aims to equip Hydro-Québec's current and future fleet of inspection robots with autonomous inspection capabilities. The three main objectives of the project are:
1. Leverage breakthroughs in artificial intelligence to enable robotic vehicles to realize real-time automated visual inspection of the company's infrastructure.
2. Facilitate and accelerate deep neural network (DNNs) machine learning through visual simulation and synthetic images.

State-of-the-art forecasting: Demand planning is a critical part of a business’ operations. Traditional approaches to forecasting use statistical methods to predict future demand from past transactions, but do not take into account contextual data. However, there are good reasons to believe that contextual data - such as weather, events, product descriptions, sentiment analysis (from reviews, Zendesk tickets, social media), and more - can contribute significant signals that directly influence forecasting accuracy for the better.

Le projet consiste à optimiser les pratiques de gestion de la publicité sur les médias sociaux (SMM) pour plusieurs clients de l’agence de Marketing Numérique Click & Mortar. Le SMM inclut plusieurs pratiques d’optimisation de la présence sur les réseaux sociaux tels que Facebook ou Instagram. Les algorithmes sophistiqués utilisés pour filtrer le contenu publié sur les réseaux sociaux oblige de plus en plus les entreprises à se tourner vers les placements publicitaires. Ceux-ci permettent aux entreprises de bénéficier d’une plus grande portée et d’un ciblage plus précis.