L’utilisation des algorithmes de type « boîte noire » gagne en popularité en entreprise, même au niveau de l’aide à la décision. En assurance collective, ces algorithmes peuvent être d’un précieux soutien pour l’actuaire chargé de la négociation avec un groupe. Toutefois, dans ce contexte, la difficulté à comprendre les prédictions de l’algorithme nuit grandement à l’appropriation de l’outil par l’actuaire. L’objectif général du projet est de trouver une méthode appropriée pour expliquer « post hoc » les prédictions de l’algorithme utilisé pour soutenir la décision de l’actuaire.
When a previously trained machine learning model is put into production, the production phase begins where said model makes predictions on the inputs provided to it. When the distribution of production data changes over time, we talk about data drift. Then the model is likely to become less efficient, or even obsolete. The project consists of building an intelligent system capable of alerting in the event of a data drift that would have a significant impact on the system.
During the last decade, we observe in organizations a surge of numbers of cyber-attacks originating internally. In this project, we aim to develop deep learning models to detect suspicious activity (such as malicious events, system failure or attacks) from log data generated by the Desjardins ecosystem.
L’un des enjeux majeurs dans le domaine de l’assurance est la fraude. Une étude américaine proposait il y a quelques années que celle-ci représentait environ 10% de chaque dollar que paie une compagnie d’assurance en dédommagement. L’incidence de la fraude est importante pour Intact sur ses résultats financiers, mais également pour l’ensemble de ses clients, car le coût de la fraude se retrouve payé dans chacune des primes d’assurance. Intact traite environ 500 000 réclamations en automobile et 150 000 réclamations en habitation chaque année.
In the last few years, a high increase in the interest of traders and investors towards financial instruments directly led to an important augmentation of the information received daily by exchanges. Exchange regulators, who constantly monitor markets to unveil potential infractions, traditionally perform their investigation manually and the notable growth in market activity represents an important risk of fraudulent events going unnoticed. In response to that new reality, exchanges around the globe are establishing automated surveillance systems that track market activity.
This project aims to use and experiment deep learning technique on modern recommender systems such as Graph Convolutional Network. The purpose of this implementation will be to drastically improve recommendation structure’s benchmark. This will allow extract user’s embedding by mapping from pre-existing features that describe the user such as ID and relevant attributes.In this project students will be integrated as a member of the advanced analytics research team that includes multiple PhD holders in relevant domains.Students would work on the following main topics: 1.
En assurance commerciale, il est important de pouvoir comprendre l'industrie dans laquelle opère une entreprise afin de bien identifier les risques auxquels l’entreprise est exposée. À cette fin, l'approche traditionnelle consiste à assigner à chaque entreprise un code d'industrie. Cependant, cette assignation est problématique car la plupart des classifications d'industrie comptent des centaines et souvent même des milliers de classes.
Car (automobile) insurance is a very common type of insurance: policyholders pay a premium to get financial compensation in case an accident happens with their cars. Insurance companies use complex calculations and a lot of information to determine the value of these premiums. More specifically, they must also consider their expectations of the future. Predicting the future is impossible but with the help of artificial intelligence, the current project aims to improve the understanding of how a portfolio of insured cars can evolve in the upcoming years.
Currently, call centres effort in this matter is largely reactive. Someone calls in, they are upset, and agents respond accordingly. However, this approach is not always most effective, especially with difficult customers. Therefore, knowing the customers current emotional state is very important for appropriate problem solving.
