This project aims at developing a conversational agent for autistic people and their families, based on highly accurate training datasets while deploying Natural Language Processing (NLP) and Deep Learning (DL) techniques. This project will help promote inclusion and diversity into AI design by using the right data to train the conversational agent system to be inclusive, while taking into consideration gender roles, age, diversity and social and ethical problems. This could eventually result in a more diverse and inclusive world.

Nowadays, software projects are no longer isolated but drive the business process of many non-IT companies. With the rise of AI applied in many industries, the problem of optimally scheduling tasks and allocating proper resources has significantly increased the challenge because of the diversity of tasks and stakeholders in the project. Due to the large volume and dynamic nature of the required information, manual optimization is typically error-prone and inefficient.

Bien que des systèmes de monitoring des patients existent déjà, leur accessibilité est souvent limitée en milieu hospitalier, réservé aux soins aigus. De plus, ils sont de nature particulièrement encombrante de par le nombre d’électrodes, capteurs et fils qui les compose. Un système de monitoring (‘wearables’) plus léger, plus accessible, et pas cher intégré aux algorithmes d’intelligence artificielle permettrait la surveillance d’un plus grand nombre de patients dans des environnements plus variés (c.-à-d.

Avec les progrès prodigieux des dernières années en matière d’agriculture de précision, ce projet apporte une solution originale à la problématique de la détection automatique des mauvaises herbes ainsi que leur éradication dans les bleuetières de bleuets nains (sauvages). Le principal objectif de ce projet consiste à développer un système robotisé autonome capable d’effectuer la détection automatique de mauvaises herbes permettant ainsi de cibler plus précisément les zones infestées au sol et ainsi orienter en 3D les opérations phytosanitaires de sarclage.

New point-of-sale (POS) machines help small businesses catalog transactions and inventory by warehousing customer, vendor, product, and sales data. This data, however, is usually warehoused in a data table that is not accessible to modern analytics and management software, such as Lightspeed. To help these businesses take advantage of their data, Enkidoo provides a service to export small business data by building an extract-transform-load (ETL) pipeline to Lightspeed. However, this process can be tedious, due to mismatches in column data and the template.

The enormous amount of data generated can be exploited using state-of-the-art AI algorithms to drive business decisions. However, a significant drawback of existing approaches is that the algorithms require a considerable amount of human effort and energy to prepare and annotate the data. Recent advances in deep learning and AI propose to solve this bottleneck using a paradigm referred to as 'Unsupervised' algorithms.

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.

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.

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.

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.