Nedbank Research Chair

I currently hold the Nedbank Research and Innovation Chair. The Chair is housed at the University of Johannesburg’s Institute for Intelligent systems.

The Chair

The University of Johannesburg’s Institute for Intelligent Systems has collaborated with Nedbank in establishing a Nedbank Research Chair. The Research Chair will provide leadership for UJ and Nedbank moving forward into the Fourth Industrial Revolution (4IR). This leadership will be nurtured through applied research and the development of the skills of the future. In developing this human capacity, the Chair will expand scientific research and innovation in areas aligned to the multi-disciplinary nature of 4IR. The Chair, as part of its activities, will seek to improve international research and innovation competitiveness in 4IR - while responding to national social and economic challenges. Generate research excellence in topics aligned with Nedbank and UJ’s vision of 4IR. The Chair will with Nedbank seeks to develop an innovative ecosystem for mid-career researchers through championing and expanding existing talent pathways. These pathways will include Master Classes and academic interventions and coordination across UJ Faculties and the College of Business and Economics.

Objectives

The Nedbank Research Chair aims to achieve its mandate through the pursuit of the following objectives:

  1. Engage in novel research projects in banking-related machine learning, artificial intelligence, meta-heuristic optimisation and data science. Results should be published and or present at conferences, workshops, and seminars and made accessible to Nedbank.
  2. Improve talent acquisition through attracting and retaining researchers and growing the Masters and Doctoral graduates.
  3. Expanding UJs course offerings at the postgraduate level to include subjects aligned with Nedbank and UJs vision of 4IR.
  4. Providing input as part of an advisory Board in conjunction with Nedbank business partners.
  5. Develop an innovative ecosystem for mid-career researchers through championing and expanding existing talent pathways.
    1. Developing an applied research and innovation culture within Nedbank. This will be achieved by increasing the number of employees actively involved in postgraduate studies in fields related to machine learning, artificial intelligence, meta-heuristic optimisation and data science.
    2. Master Classes and academic interventions and coordination across UJ Faculties and the College of Business and Economics. This includes webinars and community engagements as seen valuable to both organisations.
    3. Exploring gamification as a mechanism for championing and expanding existing talent pathways.

Research Project Themes

Privacy Protecting Machine Learning

To build every increasingly effective predictive model requires ever-increasing exploitation of personal information. However, an organisation wanting to leverage this personal information runs the legitimate risk of, implicitly or explicitly, violating the privacy of individuals. In response to this challenge, we propose to research the application of mechanisms such as homomorphic encryption in a banking environment.

Explainable Artificial Intelligence

Artificial intelligence (AI) and machine learning (ML) present opportunities for the vast improvement in almost every aspect of banking’s predictive and prescriptive automation. Unfortunately, the estimation and optimisation gains of AI and ML frequently come at the cost of interpretability and lack of transparency to the models. We propose researching how one might improve the interpretability of ML models in a banking context.

Data-driven business process automation and optimisation

Machine learning and AI, are effective techniques for automated prediction/estimations in business and industrial processes. However, they merely take you part of the way to improving decision making within an organisation. Increasingly it is no possible to automate the actions initiated as a result of the outputs of predictive models. For instance, it is often desirable to optimise some process in real-time based on the outputs of predictive models. Research into data-driven business process automation and optimisation (prescriptive analytics) would allow a bank further extract value from the already useful predicted analytics they may have.

Inference using Agent Base Modelling and Simulations

Many real-world and mathematically intractable problems, but you can still draw inferences about them using agent-based modelling and simulations (ABMSs). For instance, many economic, biological, organisation and epidemiological models can be understood in terms of ABMSs. Being able to calibrate these models to allow them to reflect real-world data remains an open challenge due to computational and intractability constraints. Using data-driven surrogate models allows us to overcome some of these challenges and researching a general understanding of the approaches remains an interesting research avenue.

Predictive analytics and Machine learning in banking

Computer Vision, Natural Language Processing, Reinforcement Learning, Unsupervised Learning, Forecasting, Anomaly Detection, Classification and Regression are all extensively valuable AI and ML techniques. These technologies can be utilised within a banking environment for numerous applications. What is not clear is which applications would benefit the most and how exactly one should go about utilising these techniques in those areas of the business. Applied research into how one might use ML and AI in specific banking applications is a pressing research area requiring further investigation.

Nedbank-Chair nedbankchair