Tabular Data Synthesis Using Generative Adversarial network

This project is about synthesizing data using generative adversarial network (GAN). Unlike conventional studies which use anonymization techniques for removing private information of individuals, we use variants of GAN architectures for crafting new records contextually like real records in the legitimate dataset. We plan to run exploratory experiments on public datasets to provide enough grounds for the viability of GANs in synthesizing information. The objective is to develop a proof of concept that shows if synthetic data could be used with similar results than original data.

Designing global event sets of floods and tropical cyclones under future climates for underwriting, capital management and regulatory purposes

With mounting pressure coming from regulators and other bodies worldwide, the financial services industry (banks, insurers, and reinsurers) will soon need to disclose and stress test their solvency and profitability to various climate scenarios. The work from the Task Force on Climate-related Financial Disclosures (TCFD) thus provides guidance as to how it should be accomplished. Physical risk assessment of the impacts of climate change remains however an important challenge for the global reinsurance industry requiring catastrophe models to be connected to climate models.

Alpha Portfolio with Reinforcement Learning - Part 2

ML/AI is widely used and deployed in many industries. Its deployment in Asset Management industry (and especially in Canadian pension fund sector) is significantly behind. Part of it is the fear of “black box” and what recommendation it gives. This sentiment is outdated as the recent advancements in ML/AI allow looking inside the “black box and thus focus on “white box” asset allocation recommendations. Another reason is that asset management these days is the intersection of three disciplines: Financial Economics, Statistics, and Computer Science.

Alpha Portfolio with Reinforcement Learning - Part 1

ML/AI is widely used and deployed in many industries. Its deployment in Asset Management industry (and especially in Canadian pension fund sector) is significantly behind. Part of it is the fear of “black box” and what recommendation it gives. This sentiment is outdated as the recent advancements in ML/AI allow looking inside the “black box and thus focus on “white box” asset allocation recommendations.
Another reason is that asset management these days is the intersection of three disciplines: Financial Economics, Statistics, and Computer Science.

Style Portfolios with Machine Learning

ML/AI is widely used and deployed in many industries. Its deployment in Asset Management industry (and especially in Canadian pension fund sector) is significantly behind. Part of it is the fear of “black box” and what recommendation it gives. This sentiment is outdated as the recent advancements in ML/AI allow looking inside the “black box and thus focus on “white box” asset allocation recommendations.
Another reason is that asset management these days is the intersection of three disciplines: Financial Economics, Statistics, and Computer Science.

Gestion des risques en cybersécurité et stratégies face à la menace des rançongiciels

Depuis quelques années un nouveau stratagème criminel s’est taillé une place de choix au sein de la communauté des cybercriminels et a pris un essor considérable : les rançongiciels ou ransomware. Ces logiciels malveillants qui infectent les ordinateurs de leurs victimes et demandent en échange le paiement d’une rançon sont devenus une des cybermenaces les plus importantes pour les entreprises en matière de cybercriminalité.

Developing rAAV-mediated retinal gene therapy to improve vision for ZellwegerSpectrum Disorder - Year Two

Peroxisome Biogenesis Disorders of the Zellweger Spectrum (PBD-ZSD) are a group of inherited genetic disorders caused by mutations in any one of 13 PEX genes. Individuals with the common PEX1-G843D mutation consistently develop a retinopathy that progresses to blindness. To test whether we could slow visual loss in these patients, we performed a proof-of-concept trial for PEX1 retinal gene augmentation therapy using our mouse model homozygous for the equivalent PEX1-G844D mutation.

Intact – Analyse et extraction de caractéristiques de voitures à partir d’images

Intact Corporation financière est le plus important fournisseur d'assurances multirisques au Canada en primes annuelles.
Intact vise à offrir un service de réclamations accéléré à ses clients. Au moment d’ouvrir une réclamation, Intact demande d’ores et déjà à ses clients de fournir des images du véhicule qui permettent d’identifier préalablement la condition générale du véhicule. Il sera demandé au client de mettre en évidence la pièce d’équipement endommagée, en s’assurant que celle-ci est bien visible dans l’image.

Simulation-based decision support system for data analytics deployment

Data has been recognized as one of the most valuable assets of modern business. The capacity to gather, store, analyze and interpret data in great quantities can determine to a large degree the ability of a company to achieve goals and adapt to largely volatile environments. This is especially true for financial institutions where data is directly connected to profitability.

Data synthesis using generative adversarial network

This project is about synthesizing data using generative adversarial network (GAN). Unlike conventional studies which use anonymization techniques for removing private information of individuals, we use variants of GAN architectures for crafting new records contextually similar to real records in the legitimate dataset. We plan to run exploratory experiments on public datasets to provide enough grounds for the viability of GANs in synthesizing information. The objective is to develop a proof of concept that shows if synthetic data could be used with similar results than original data.

Pages