Intelligence Artificielle et Apprentissage Machine pour la Cybersécurité

Ce projet consiste à extraire la nature des accès informatiques des employés d’une organisation et les consolider afin de permettre au gestionnaire de l’employer ou à un auditeur d’avoir une vue d’ensemble des accès informatiques de ses employés.

Causal Recommender Systems for Sequential Decision-Making

Recommender systems (RS) are intended to be a personalized decision support tool, where decisions can take the form of products to buy (e.g., Amazon), movies to watch (e.g., Netflix), online news to read (e.g., Google News), or even individuals to screen for a medical condition (e.g., personalized medicine). For digital users, RS play an essential role, since the available content (and hence possible actions) grows exponentially.

Lifelong reinforcement learning with autonomous inference of subtask dependencies

In this project, we propose a continual learning approach to face the problem of forward transfer in complex reinforcement learning tasks. Concretely, we propose a model that learns how to combine a series of general modules in a deep learning architecture, so that generalization emerges through the composition of those modules. This is of vital importance for Element AI to provide reusable solutions that scale with new data, without the need of learning a new model for every problem and improving the overall performance.

Application of Deep Learning techniques in stock ranking for different horizon returns

One of the approaches portfolio managers commonly use to build portfolios, is to rank the underlying assets based on the prediction for the stock returns, as well as other aspects of the portfolio such as the portfolio risk. In this project we aim to apply different deep learning techniques to the problem of stock ranking. The features we want to use to train our models are mainly derived from fundamental company data including quarterly and annual filings of the publicly traded companies.

Blockchain-based secure transaction logging system

Smart-city platform developed by B-CITI company stores sensitive information of platform users (citizens) in a database to provide them services. The proposed research project’s aim is to build a system that records each user interaction with the database. The records are immutable and, easily linkable and traceable. The records can be used as forensics evidence for proving and validating the integrity of the data in the database. For this, we use permissioned blockchain that is managed by multiple untrusted entities to generate trust in the system.

Semantic versioning of model changes in decision support systems

Nowadays, almost any company in Canada in operation heavily relies on software solutions to improve their productivity. However, they are often facing the problem of having too many options to choose from for the software best fit for their needs. Decision support systems (DSS) help enterprises to take significant business decisions, such as finding the best software solution. Our industrial partner has developed a DSP that incrementally builds a decision model with customers preferences, choices, and ratings.

Learning priors for data-efficient causal discovery

The inference of causal relationships is a problem of fundamental interest in science. Compared to models that rely on mere correlations, causal models allow us to anticipate the effect of a change in a system. Such causal models have applications ranging from government policy making to personalized medicine. However, learning causal models from data is a challenging task, since it requires large data sets and, in some cases, the conduction of costly or invasive experiments. In this project, we propose a new method to learn causal models using less data.

Recommender System for Game-Like Machine Learning Systems

Machine learning (ML) offers a tremendous opportunity for developing new algorithms in various field of activities. Taking image classification for instance, usually the best practices are found from trial and errors and usually taking several approaches and compare one versus another. However, it is not obvious for a non-expert, and even for an expert, what ML pipeline should work best on a given dataset.

Minimiser les efforts d’annotation lors du développement d’un modèle prédictif en traitement des langues – Phase 1

Développer un modèle prédictif en traitement automatique des langues requière la création d’un corpus annoté : un texte et des annotations que l’on tentera de reproduire automatiquement. Il s’agit d’une activité à la fois complexe (les annotations sont souvent du ressort d’un expert) et coûteuse (annotations méticuleuses à produire en grande quantité). Le projet vise à développer une expertise pour minimiser les interventions (annotations) permettant d’obtenir un modèle prédictif d’une qualité donnée.

Implementation and Characterization of a Prototype of Optical Free-space Interconnect for Space Applications

As we advanced into the information age, the need for high capacity communication channels is becoming ubiquitous. As the next generation of satellites is being deployed there is a need for efficient interconnections between them, especially for those forming low earth orbiting constellations for the coming Internet-of-Space applications. The limited radio frequency spectrum available is not sufficient to implement these communication links, and thus, free-space optical interconnects (FSOIs) are expected to become the technology of choice to interconnect satellites.