Scaling Security Architecture

Security has become increasingly important as cyberattacks are more prominent. Despite advances in security technologies there's still a need for a new security architecture that is simply adaptable and scalable to different and evolving security requirements of various organizations. This project will study the latest technologies and solutions in the field of Cybersecurity that are being used to design scalable cybersecurity frameworks. We will also look at the new technologies and processes that are disrupting the existing Cybersecurity landscape.

Drivers of Time to Resolution, Application of LASSO Regression and Random Forest

International Financial Reporting Standards (IFRS) for loss allowances are changing, and financial institutions are proactively adapting existing methodologies and developing new ones to remain compliant. The main ingredient in the myriad of evaluations that banks are required to perform for compliance is risk assessment. The first goal of this research project is to review best practice risk models, with a special focus on modeling the evolution of default probabilities.

Dynamic Credit Scoring

Banks use a myriad of methodologies to inform their officers on credit extension decisions. One of the most employed approach is to summarize borrower creditworthiness by credit scores, which in turn depend on loan default probabilities. The probability of default depends both on borrower characteristic and on the overall state of the economy. The goal of this project is to create credit scores that are responsive to the expected state of the economy.

Forecasting Ability of Non-consumer Scorecards and their Ability to Predict Probability of Default

International Financial Reporting Standards (IFRS) for loss allowances are changing, and financial institutions are proactively adapting existing methodologies and developing new ones to remain compliant. The main ingredient in the myriad of evaluations that banks are required to perform for compliance is risk assessment. The first goal of this research project is to review best practice risk models, with a special focus on modeling the evolution of default probabilities.

Affine Multivariate GARCH Models

The objective of the proposed research program is to develop a flexible and unified multivariate framework for modeling the returns of financial assets. The program is innovative since it establishes closed-form formulas for an efficient and reliable calculation of risk measures and derivative prices. For financial institutions and government regulators, who are performing pricing and risk management calculations very frequently with thousands of assets, closed form solutions are of immense importance.

Statistical machine learning methods applied to ATB data for debt collection optimization, small business lending decision modelling, and open banking initiatives

The intern will research new modelling technology to determine if the new models can make a significant improvement in servicing customers for loan approvals, debt collections, and open banking. The intern will work closely with the partner to understand the banking process and opportunity. The partner organization will receive several benefits from working with the innovative and knowledgeable intern including cross-training of techniques through collaboration, enhanced model accuracy, and enabling the partner to test new techniques.

Identifying the Key Factors in Adopting Digital ID Service and Profiling the Early Adopters

Despite the digital transformation going on around us, the methods of proving our identity remain locked in a traditional physical mode, with paper or plastic documents, and the reliance on these traditional modes of proving identification and authentication is becoming a significant barrier to innovation. Therefore, creating a digital identity and authentication method that is highly secure, ubiquitous and convenient is necessary. A new digital identity system will allow us to do things online that we have traditionally used physical ID for.

A reconciliation of the top-down and bottom-up approaches to risk capital allocations

Two overarching approaches to allocate the aggregate risk capital stand out nowadays. These are the top-down approach that entails that the allocation exercise is imposed by the corporate centre, and the bottom-up approach that implies that the allocation of the aggregate risk to business units is informed by these units. Briefly, the top-down allocations start with the aggregate risk capital that is then replenished among business units according to the views of the centre, thus limiting the inputs from the business units.

Modeling Exfiltration Events in Sunlife Cybersecurity Data

Many governments and other organizations hold confidential data. Theft of that data can be extremely damaging both to the organization and to the people whose data has been stolen. Massive breaches each involving millions of people have been occurring on a regular basis in recent years. New Cyber Security tools are needed to help people determine the threats that exist and to provide active online monitoring that can detect unusual behavior as it happens.

Fast and Accurate Computation of Wasserstein Adversarial Examples

Machine learning (ML) has recently achieved impressive success in many applications. As ML starts to penetrate into safety-critical domains, security/robustness concerns on ML systems have received lots of attention lately. Very surprisingly, recent work has shown that current ML models are vulnerable to adversarial attacks, e.g. by perturbing the input slightly ML models can be manipulated to output completely unexpected results. Many attack and defence algorithms have been developed in the field under the convenient but questionable Lp attack model.

Pages