When the opportunity presented itself for MultiChoice to partner with the University of Pretoria five years ago, the company had no idea that the benefits would exceed its expectations to the degree that it has. The partnership entailed the development of scarce artificial intelligence (AI) skills − specifically as they relate to machine learning (ML) technology − in South Africa.

In 2018, when the idea of a Research Chair in Machine Learning at the University of Pretoria was born, Koos Bekker, the Chairperson of Multichoice’s parent company at the time, the Naspers Group, had already experienced the cutting-edge research that was being conducted in the University’s Faculty of Engineering, Built Environment and Information Technology.

By that time, a circuit chip that had been developed by researchers in the Carl and Emily Fuchs Institute for Microelectronics (CEFIM) in the University’s Department of Electrical, Electronic and Computer Engineering, was already in use in the company’s MNet decoders.

The management of MultiChoice was so impressed by what they had seen at the University of Pretoria in terms of research and human capital development, that discussions ensued with the Faculty’s former Dean, Prof Sunil Maharaj (now Vice-Principal: Research, Innovation and Postgraduate Education). These discussions would lead to collaboration between the two entities to establish a mutually agreed roadmap that would result in the establishment of a Research Chair in Machine Learning.

At that stage, machine learning was just starting to gain traction as an emerging discipline in computer science and computer engineering. Companies in the information and communication technology (ICT) sector were recognising the benefits of investing in these disruptive skills.

The University’s Faculty of Engineering, Built Environment and Information Technology is one of the foremost faculties of its kind in Africa. Its programmes in Computer Science and Information Systems are ranked in the top 550 in the world according to the latest QS World University Rankings by Subject in the field of engineering and technology.

MultiChoice therefore considered a partnership with the University to acquire and identify human resource capacity for the company, which would result in a win-win engagement. According to the Memorandum of Agreement, the University would provide MultiChoice with human capital development, research and development, consultancy services, advisory, supervisory and managerial support, as well as resources that could contribute to the development of specific projects to the benefit of MultiChoice. The agreement included the awarding of bursaries to promising students in computer science and computer engineering.

MultiChoice would be granted access to the creative projects that were developed during the students’ research. In addition, MultiChoice grants students the option of completing its vacation work programme, as well as the opportunity of employment upon conclusion of their studies.

Further to supporting students financially, MultiChoice also provides guidance in terms of potential research topics that could benefit the University in terms of research output, as well as MultiChoice in terms of maintaining its competitive edge in the industry.

Some current and future topics include the following:

• Technical failure analysis
• Real-time monitoring
• The extraction or creation of metadata
• Video-on-demand quality assurance
• Cloud resource allocation
• Machine learning on embedded systems
• Social media analytics
• Customer profiling for automated testing
• Computer network intrusion detection
• Piracy monitoring
• Forecasting weather-related impacts on service delivery

The outputs of the research emanating from the projects associated with the Chair in Machine Learning not only benefit the University and MultiChoice, but also have the advantage of creating more knowledge and competence in South Africa. The ability to harness machine learning has seen some excellent projects from both the computer science and computer engineering domains. A student project aimed at automatic thumbnail selection using machine learning, for example, has been accepted for deployment at MultiChoice. This is a very exciting outcome of the partnership.

Another project, which focused on the development of a ML-based recommender system, is also being implemented by MultiChoice. Following the conclusion of their studies, two graduates of the Computer Engineering programme have been appointed by MultiChoice, with a third graduate joining them before the end of the year. These graduates not only illustrate the success of the programme, but also the value of employing a graduate from the University of Pretoria. As employees, they are truly phenomenal – they are well-rounded individuals, who fit into the context and culture of the organisation right away. They are able to step into their role in the company with an immediate understanding of what is expected of them and how to work as a member of a team.

I have a great appreciation for the role played by the two chairholders at the University of Pretoria: Prof Nelishia Pillay (Computer Science) and Prof Pieter de Villiers (Computer Engineering). Their supervision and guidance of students at undergraduate and postgraduate level have definitely contributed to the success of the programme.

I look forward to partnering with the University of Pretoria for an additional five-year term, and to see the current projects reaching their anticipated conclusion to the benefit of the country at large.

Gerdus van Eeden
CTO: Multichoice