Master of Science in Machine Learning Engineering

Purpose

The purpose of the Master of Science in Machine Learning Engineering qualification is to develop postgraduate learners with high-level research abilities, and a good knowledge base in the field of Machine Learning Engineering. Qualifying learners are prepared for work in the field of Machine Learning Engineering as well as (depending on additional skills) Data Science and Data Engineering. In this manner, the qualification will further skills development in a specialist area, serve to assist industry professionals in upskilling and diminish the skills shortage within the industry. It will also seek to improve the self-learning capabilities of each learner to promote lifelong learning and an aptitude for developing other learners in similar fields.

Upon completion of this qualification, qualifying learners will be able to

• Identify and accurately analyse problems within the Machine Learning Engineering framework and environment by researching problems creatively and innovatively.
• Organize and manage activities responsibly, effectively, and ethically, accept and take responsibility within his or her limits of competence, and exercise judgement based on knowledge and expertise, pertaining to the field of research.
• Plan and conduct applicable levels of investigation, research and or experiments by applying appropriate theories, methodologies, and interpretation.

Rationale

The world is inundated with data, constantly moving, and renewing itself. Such vast amounts of data are of no use if there are no people able to collect, collate, analyse, and apply it to the benefit of society. This is where a qualification in Machine Learning Engineering becomes a fundamental stepping stone to address South Africa's growing need for scientists and engineers capable of utilising data to its full. To obtain insight from data in various forms with various outcomes, both structured and unstructured, various scientific components need to be collated and structured within an often-existing technological framework. As such, it is not only necessary to design the relevant machine learning models but also to ensure that the pipelines and structures are in place to bring such a model into production. Only in this manner can the true benefit of data science and machine learning be realised.

Machine Learning Engineering is a multi-disciplinary field, involving a wide range of disciplines, from applied mathematics to statistics and machine learning to software engineering, which places it within the realm of the University's vision for driving the 4IR. Advances in computer technology and processing speed, the relatively low cost of storing data, and the massive availability of data from the Internet and other sources have provided the ideal platforms for taking data and making meaning from it for the benefit of society.

The field of Machine Learning Engineering by its very nature requires highly skilled individuals. The level of scientific expertise along with the practical experience of engaging with the technicalities of application, require such individuals to straddle both academia and industry. As such, effective research entities are often ideal to host qualifications in such a field, given their collaborative mindset with regard to research endeavours, and engagement with industry. The latter provides the opportunity to source projects from industry, which often leads to highly impactful and practical research.

Furthermore, given the high-level, unique, and diverse skills needed by Machine Learning Engineers, the demand for such individuals has grown exponentially, far outstripping the current supply.

Globally and locally, the education system must be empowered to design relevant qualifications often context-specific, given the nature of the requirements within a certain company to stay abreast of this demand. The industry does not have the capacity or necessary skill at times to conduct the required training without the support of formal higher education institutions. As such, the only way the next generation of Machine Learning Engineers, Data Scientists and Data Engineers can be taught is through an appropriate and effective postgraduate academic qualification, designed to consider the need for part-time, distance study.

The development of this qualification in particular would further allow for the development of cutting-edge research in Machine Learning Engineering, Data Science and Data Engineering. This qualification would be part of developing the next generation of scientists or engineers capable of converting data intelligence into business value for any industry. Qualifying learners will have the ability to tackle problems across social, economic, and technical fields, allowing them to apply a cross-disciplinary and complex lens to the multi-layered challenges of the 21st century.

The impact of qualified professionals in this area will be immeasurable. This will be a flagship qualification, leading the way into a data-driven society where individuals with relevant skills are of key importance. There is much demand in the industry such as, in finance, telecommunications, agriculture, health etc. as well as for the upskilling of staff around Machine Learning Engineering, Data Science and Data Engineering. Given the wide scope of the field, being a Machine Learning Engineer covers a range of professions including, but not limited to, engineers, computer scientists, applied mathematicians, physicists, and machine learners.

The qualification is aimed at learners with a Bachelors Honours Degree or any related qualification at NQF Level 8 with mathematical background, and in particular:

• Individuals who have a data-related background and want to transition.
• Individuals who have a technical programming background and want to focus on data-related problems.
• Learners with degrees in Mathematical Sciences/Engineering (BSc/BEng)
• Data Engineers (Emerging Machine Learning Engineers)
• Data Scientists (Upskilling their understanding of Machine Learning Engineering)
• Machine Learning Engineers who require a deeper understanding.

The Research Group, Data Across Disciplines, which is driving this qualification is comprised of a group of individuals who are all experts in appropriate fields. They are active in research, teaching, and postgraduate supervision, and have links with international and local collaborators in academia and industry. These partnerships will provide the prospects for the sustainability of the proposed qualification. Furthermore, the qualification has been structured in a way which will allow it to speak to and support future qualifications in Data Science and Data Engineering for instance, and allows articulation between a variety of Honours and a Master's qualifications in Science, the Academy of Computer Science, and the School of Electrical Engineering at the very least.

Recognition of Prior Learning (RPL)

The institution has an approved Recognition of Prior Learning (RPL) policy applicable to equivalent qualifications for admission into the qualification. RPL will be applied to accommodate applicants who qualify. RPL thus provides alternative access and admission to qualifications, as well as advancement within qualifications.

RPL for access

• Learners who do not meet the minimum entrance requirements or the required qualification that is at the same NQF level as the qualification required for admission may be considered for admission through RPL.
• To be considered for admission in the qualification based on RPL, applicants should provide evidence in the form of a portfolio that demonstrates that they have acquired the relevant knowledge, skills, and competencies through formal, non-formal and/or informal learning to cope with the qualification expectations.

RPL for exemption of modules

• Learners may apply for RPL to be exempted from modules that form part of the qualification. For a learner to be exempted from a module, the learner needs to provide sufficient evidence in the form of a portfolio that demonstrates that competency was achieved for the learning outcomes that are equivalent to the learning outcomes of the module.

RPL for credit

• Learners may also apply for RPL for credit for or towards the qualification, in which they must provide evidence in the form of a portfolio that demonstrates prior learning through formal, non-formal and/or informal learning to obtain credits towards the qualification.
• Credit shall be appropriate to the context in which it is awarded and accepted.

Entry Requirements

The minimum entry requirement for this qualification is

• Postgraduate Diploma in Artificial Intelligence, NQF Level 8.

Or

• Bachelor of Commerce Honours in Mathematical Statistics, NQF Level 8.

Or

• Bachelor of Commerce Honours in Statistics, NQF Level 8.

Or

• Bachelor of Engineering in Electrical Engineering, NQF Level 8.

This qualification consists of the following compulsory and elective modules at NQF Level 9 totalling 165 Credits.

Compulsory Modules, Level 9, 150 Credits

• Applied Machine Learning, 15 Credits.
• Statistics and Data Analysis for Engineers, 15 Credits.
• Probabilistic Graphical Models, 15 Credits.
• Convex Optimizations for Machine Learning, 15 Credits.
• Machine Learning System Design, 15 Credits.
• Interpretable Machine Learning, 15 Credits.
• Research Project in Machine Learning Engineering, 60 Credits.

Elective Modules, Level 9, 15 Credits: (Choose one of the following modules)

• Special Topics in Engineering Applications of Machine Learning, 15 Credits.
• Topics in Advanced Machine Learning, 15 Credits.
• Large Scale Machine Learning, 15 Credits.
• Machine Learning for Multimedia Signal Processing, 15 Credits.
• Machine Learning for Signal and Image Processing, 15 Credits.
• Machine Learning for Optimal Control, 15 Credits.

1. Identify and accurately analyse problems within the Machine Learning Engineering framework and environment by researching problems creatively and innovatively.
2. Organise and manage activities responsibly, effectively, and ethically accept and take responsibility within competence, and exercise judgement based on knowledge and expertise, pertaining to the field of research.
3. Plan and conduct applicable levels of investigation, research and or experiments by applying appropriate theories, methodologies, and interpretation.
4. Communicate effectively, both orally and in writing, with specific research audiences and the community at large, using appropriate methodologies and interpretation.
5. Demonstrate applicable cultural, and aesthetic sensitivity across a range of social and environmental contexts in executing Machine Learning Engineering research or development activities.
6. Demonstrate an appropriate understanding of the relevant topics which provide the knowledge base of an expert Machine Learning Engineer.

Associated Assessment Criteria

Associated Assessment Criteria for Exit Level Outcome 1

• Correctly identify and assess machine learning tools and isolate merits and the circumstances for most appropriate and efficient use.
• Illustrate the effective application and integration of knowledge within the solution design for a real-world application.
• Cohesively link findings to applicable theoretical underpinnings, as well as analyse outputs from their algorithms and problem designs.

Associated Assessment Criteria for Exit Level Outcome 2

• Complete projects and mini dissertation timeously, ethically, and responsibly.
• Apply good judgement and be cognizant of the context within which various applications will be studied and engaged to inform appropriate conduct.
• Work with minimal supervision, be able to isolate quality sources of information and synthesize this information appropriately to support their work.

Associated Assessment Criteria for Exit Level Outcome3

• Distinguish between appropriate methodologies to consider during the coursework and the research investigations.
• Offer relevant theoretical underpinnings for the arguments.
• Accurately interpret and analyse the data gathered, the quality of the software or algorithms designed, as well as the output obtained from such algorithms.

Associated Assessment Criteria for Exit Level Outcome 4

• Articulate findings using appropriate methodologies and interpretation.
• Coherently communicate with various stakeholders during all phases of the study.
• Integrate theoretical underpinnings with the data gathered, algorithms designed and generated output in a cohesive manner.

Associated Assessment Criteria for Exit Level Outcome 5

• Illustrate an awareness of the social impact of research, especially within the context of data usage.
• Obtain the required ethical clearance necessary for research.
• Illustrate an awareness of the environmental impact of research.
• Comply with Health and Safety legislation during research activities.

Associated Assessment Criteria for Exit Level Outcome 6

• Identify and apply various algorithms.
• Use data analysis techniques, and software technologies required to perform well as Machine Learning Engineers.
• Implement the necessary algorithms and analyses bearing in mind best practices as well as software limitations and industry requirements.

Associated Assessment Criteria for Exit Level Outcome 1

• Correctly identify and assess machine learning tools and isolate merits and the circumstances for most appropriate and efficient use.
• Illustrate the effective application and integration of knowledge within the solution design for a real-world application.
• Cohesively link findings to applicable theoretical underpinnings, as well as analyse outputs from their algorithms and problem designs.

Associated Assessment Criteria for Exit Level Outcome 2

• Complete projects and mini dissertation timeously, ethically, and responsibly.
• Apply good judgement and be cognizant of the context within which various applications will be studied and engaged to inform appropriate conduct.
• Work with minimal supervision, be able to isolate quality sources of information and synthesize this information appropriately to support their work.

Associated Assessment Criteria for Exit Level Outcome3

• Distinguish between appropriate methodologies to consider during the coursework and the research investigations.
• Offer relevant theoretical underpinnings for the arguments.
• Accurately interpret and analyse the data gathered, the quality of the software or algorithms designed, as well as the output obtained from such algorithms.

Associated Assessment Criteria for Exit Level Outcome 4

• Articulate findings using appropriate methodologies and interpretation.
• Coherently communicate with various stakeholders during all phases of the study.
• Integrate theoretical underpinnings with the data gathered, algorithms designed and generated output in a cohesive manner.

Associated Assessment Criteria for Exit Level Outcome 5

• Illustrate an awareness of the social impact of research, especially within the context of data usage.
• Obtain the required ethical clearance necessary for research.
• Illustrate an awareness of the environmental impact of research.
• Comply with Health and Safety legislation during research activities.

Associated Assessment Criteria for Exit Level Outcome 6

• Identify and apply various algorithms.
• Use data analysis techniques, and software technologies required to perform well as Machine Learning Engineers.
• Implement the necessary algorithms and analyses bearing in mind best practices as well as software limitations and industry requirements.