Master of Science in AI and Machine Learning
Investigate an artificial-intelligence problem through advanced study and a supervised research project.
About the course
Investigate an artificial-intelligence problem through advanced study and a supervised research project. The proposed masters outline emphasises a defensible research question, credible baselines and careful evaluation. Southern African applications should consider language, data availability, representativeness and practical use, with conclusions restricted to what the experiments support.
What you'll learn
Critically evaluate research in a selected AI field.
Design a reproducible investigation with suitable baselines and evaluation methods.
Analyse errors, uncertainty and ethical implications.
Defend an independent research contribution and its practical limitations.
Requirements
Recommended preparation is a relevant recognised undergraduate qualification and readiness for academic reading, applied analysis and independent study. A masters route also requires adequate research preparation and an appropriate supervisor. BMIT must confirm admission equivalence, bridging requirements and its approved research and assessment regulations.
Course content
The subjects are grouped by learning stage, from foundations to specialist study and supervised application. Complete earlier foundations before the related advanced work. Clinical, laboratory, field and workplace subjects use approved facilities and supervision appropriate to the programme.
3 learning stages 11 subjects
1Stage 1: Postgraduate disciplinary study4 subjects
2Stage 2: Specialist analysis and application4 subjects
3Stage 3: Supervised research or applied integration3 subjects
Subjects in this programme
The subjects are grouped by learning stage, from foundations to specialist study and supervised application. Complete earlier foundations before the related advanced work. Clinical, laboratory, field and workplace subjects use approved facilities and supervision appropriate to the programme.
11 subjects are listed below with a short description of each. The Subject descriptions tab contains the learning outcomes, topics, practical tasks and assessments.
Stage 1: Postgraduate disciplinary study
- Probability and Statistical Modelling. Examine random variables, distributions and their relationship within probability and statistical modelling. The subject develops conditional probability and likelihood, then examines bayesian reasoning and model checking. Compare statistical models using a transparent teaching dataset.
- Optimisation Methods. Develop your understanding of optimisation methods through objective functions and constraints. The subject develops gradient methods and linear programming, then examines convexity concepts and sensitivity. Solve a constrained allocation or model-fitting problem.
- Machine Learning. Study learning tasks and features as foundations for machine learning. The subject develops training and generalisation, then examines model selection and error analysis. Compare a baseline and candidate model on appropriately separated data.
- Deep Learning. Examine neural representations, optimisation and their relationship within deep learning. The subject develops regularisation and model architectures, then examines training diagnostics and evaluation. Investigate a small model using an approved public dataset.
Stage 2: Specialist analysis and application
- Knowledge Representation and Reasoning. Develop your understanding of knowledge representation and reasoning through logic and rules. The subject develops ontologies and inference, then examines uncertainty and knowledge quality. Build a small knowledge model and test it against defined questions.
- Natural Language Processing. Study text representation and linguistic variation as foundations for natural language processing. The subject develops language modelling and classification, then examines translation evaluation and language-data ethics. Analyse an authorised language dataset, including relevant regional variation.
- Computer Vision. Examine image representation, features and their relationship within computer vision. The subject develops recognition and detection, then examines image evaluation and dataset bias. Compare image-analysis methods on an authorised teaching dataset.
- Responsible AI and Data Governance. Develop your understanding of responsible ai and data governance through consent and privacy. The subject develops bias and explainability, then examines human oversight and deployment accountability. Audit a proposed AI use case for data and decision risks.
Stage 3: Supervised research or applied integration
- Advanced Research Design. Study research positioning and theoretical frameworks as foundations for advanced research design. The subject develops method comparison and validity, then examines research ethics and feasibility. Defend an advanced study design against plausible alternatives.
- Research Proposal and Protocol. Examine problem significance, literature synthesis and their relationship within research proposal and protocol. The subject develops research questions and protocol development, then examines ethics approval and work plan. Develop a supervised research protocol and defend its feasibility.
- Dissertation Research and Defence. Develop your understanding of dissertation research and defence through data stewardship and analysis. The subject develops interpretation and research contribution, then examines limitations and scholarly communication. Complete approved research under supervision and retain reproducible evidence.
Assessment and practical learning
- Proposed assessment: critical literature reviews, analytical assignments and seminars which defend the choice of methods and evidence.
- An applied project or research proposal addressing a defined Southern African institutional or community problem.
- An independently assessed report or dissertation with an oral presentation, according to an approved BMIT route and assessment policy.
Practical application
Research requires appropriate computing access, an academic supervisor and permission to use the selected data. Maintain versioned code, data documentation and experimental records. Personal or sensitive data needs formal review. Claims about progress should distinguish measured performance from assumptions about deployment or generalisation.
Subject descriptions
Select a subject to read its learning outcomes, main topics, practical task and assessment.
Stage 1: Postgraduate disciplinary study
Probability and Statistical Modelling
Examine random variables, distributions and their relationship within probability and statistical modelling. The subject develops conditional probability and likelihood, then examines bayesian reasoning and model checking. Compare statistical models using a transparent teaching dataset.
Learning outcomes
- Explain random variables and distributions using an appropriate example.
- Analyse a subject-related problem involving conditional probability and likelihood.
- Present reasoned evidence addressing bayesian reasoning and model checking.
Main topics
- Random variables
- Distributions
- Conditional probability
- Likelihood
- Bayesian reasoning
- Model checking
Practical task
Compare statistical models using a transparent teaching dataset.
Assessment
Submit calculations, model checks and an interpretation.
Optimisation Methods
Develop your understanding of optimisation methods through objective functions and constraints. The subject develops gradient methods and linear programming, then examines convexity concepts and sensitivity. Solve a constrained allocation or model-fitting problem.
Learning outcomes
- Explain objective functions and constraints using an appropriate example.
- Analyse a subject-related problem involving gradient methods and linear programming.
- Present reasoned evidence addressing convexity concepts and sensitivity.
Main topics
- Objective functions
- Constraints
- Gradient methods
- Linear programming
- Convexity concepts
- Sensitivity
Practical task
Solve a constrained allocation or model-fitting problem.
Assessment
Submit a reproducible solution and sensitivity discussion.
Machine Learning
Study learning tasks and features as foundations for machine learning. The subject develops training and generalisation, then examines model selection and error analysis. Compare a baseline and candidate model on appropriately separated data.
Learning outcomes
- Explain learning tasks and features using an appropriate example.
- Analyse a subject-related problem involving training and generalisation.
- Present reasoned evidence addressing model selection and error analysis.
Main topics
- Learning tasks
- Features
- Training
- Generalisation
- Model selection
- Error analysis
Practical task
Compare a baseline and candidate model on appropriately separated data.
Assessment
Submit reproducible experiments and an error analysis.
Deep Learning
Examine neural representations, optimisation and their relationship within deep learning. The subject develops regularisation and model architectures, then examines training diagnostics and evaluation. Investigate a small model using an approved public dataset.
Learning outcomes
- Explain neural representations and optimisation using an appropriate example.
- Analyse a subject-related problem involving regularisation and model architectures.
- Present reasoned evidence addressing training diagnostics and evaluation.
Main topics
- Neural representations
- Optimisation
- Regularisation
- Model architectures
- Training diagnostics
- Evaluation
Practical task
Investigate a small model using an approved public dataset.
Assessment
Submit code, learning curves and a reasoned model comparison.
Stage 2: Specialist analysis and application
Knowledge Representation and Reasoning
Develop your understanding of knowledge representation and reasoning through logic and rules. The subject develops ontologies and inference, then examines uncertainty and knowledge quality. Build a small knowledge model and test it against defined questions.
Learning outcomes
- Explain logic and rules using an appropriate example.
- Analyse a subject-related problem involving ontologies and inference.
- Present reasoned evidence addressing uncertainty and knowledge quality.
Main topics
- Logic
- Rules
- Ontologies
- Inference
- Uncertainty
- Knowledge quality
Practical task
Build a small knowledge model and test it against defined questions.
Assessment
Submit the model, test cases and limitations.
Natural Language Processing
Study text representation and linguistic variation as foundations for natural language processing. The subject develops language modelling and classification, then examines translation evaluation and language-data ethics. Analyse an authorised language dataset, including relevant regional variation.
Learning outcomes
- Explain text representation and linguistic variation using an appropriate example.
- Analyse a subject-related problem involving language modelling and classification.
- Present reasoned evidence addressing translation evaluation and language-data ethics.
Main topics
- Text representation
- Linguistic variation
- Language modelling
- Classification
- Translation evaluation
- Language-data ethics
Practical task
Analyse an authorised language dataset, including relevant regional variation.
Assessment
Submit an evaluated language-processing project with documented errors.
Computer Vision
Examine image representation, features and their relationship within computer vision. The subject develops recognition and detection, then examines image evaluation and dataset bias. Compare image-analysis methods on an authorised teaching dataset.
Learning outcomes
- Explain image representation and features using an appropriate example.
- Analyse a subject-related problem involving recognition and detection.
- Present reasoned evidence addressing image evaluation and dataset bias.
Main topics
- Image representation
- Features
- Recognition
- Detection
- Image evaluation
- Dataset bias
Practical task
Compare image-analysis methods on an authorised teaching dataset.
Assessment
Submit experiments and an error-focused visual report.
Responsible AI and Data Governance
Develop your understanding of responsible ai and data governance through consent and privacy. The subject develops bias and explainability, then examines human oversight and deployment accountability. Audit a proposed AI use case for data and decision risks.
Learning outcomes
- Explain consent and privacy using an appropriate example.
- Analyse a subject-related problem involving bias and explainability.
- Present reasoned evidence addressing human oversight and deployment accountability.
Main topics
- Consent
- Privacy
- Bias
- Explainability
- Human oversight
- Deployment accountability
Practical task
Audit a proposed AI use case for data and decision risks.
Assessment
Submit a documented governance and evaluation plan.
Stage 3: Supervised research or applied integration
Advanced Research Design
Study research positioning and theoretical frameworks as foundations for advanced research design. The subject develops method comparison and validity, then examines research ethics and feasibility. Defend an advanced study design against plausible alternatives.
Learning outcomes
- Explain research positioning and theoretical frameworks using an appropriate example.
- Analyse a subject-related problem involving method comparison and validity.
- Present reasoned evidence addressing research ethics and feasibility.
Main topics
- Research positioning
- Theoretical frameworks
- Method comparison
- Validity
- Research ethics
- Feasibility
Practical task
Defend an advanced study design against plausible alternatives.
Assessment
Submit a research protocol and critical methodological defence.
Research Proposal and Protocol
Examine problem significance, literature synthesis and their relationship within research proposal and protocol. The subject develops research questions and protocol development, then examines ethics approval and work plan. Develop a supervised research protocol and defend its feasibility.
Learning outcomes
- Explain problem significance and literature synthesis using an appropriate example.
- Analyse a subject-related problem involving research questions and protocol development.
- Present reasoned evidence addressing ethics approval and work plan.
Main topics
- Problem significance
- Literature synthesis
- Research questions
- Protocol development
- Ethics approval
- Work plan
Practical task
Develop a supervised research protocol and defend its feasibility.
Assessment
Submit the approved proposal and a research seminar presentation.
Dissertation Research and Defence
Develop your understanding of dissertation research and defence through data stewardship and analysis. The subject develops interpretation and research contribution, then examines limitations and scholarly communication. Complete approved research under supervision and retain reproducible evidence.
Learning outcomes
- Explain data stewardship and analysis using an appropriate example.
- Analyse a subject-related problem involving interpretation and research contribution.
- Present reasoned evidence addressing limitations and scholarly communication.
Main topics
- Data stewardship
- Analysis
- Interpretation
- Research contribution
- Limitations
- Scholarly communication
Practical task
Complete approved research under supervision and retain reproducible evidence.
Assessment
Submit a dissertation and defend methods, findings and limitations orally.
Instructors
