-
RAIL
The RAIL Lab at the University of the Witwatersrand is a research group focusing on Artificial Intelligence, Machine Learning, and Robotics.The group's major focus is Reinforcement Learning (RL), both in terms of contributing the general development of the field as well as its applicability to a number of problem areas.
Other research areas include computer vision and theory of deep networks. Some of the application areas are healthcare and education.
Director: Prof. Benjamin Rosman
Deputy Director: Prof. Steven James -
PRIME Lab
The PRIME Lab at the University of the Witwatersrand is a research group focused on Machine Learning and Signal Processing. The group’s major focus is on Computer Vision, Natural Language Processing and Medical Machine Learning. Specifically, the lab focuses on developing machine learning and artificial intelligence systems for resource-constrained settings with limited data availability. Noteworthy applications include healthcare and edge devices.
Director: Prof. Richard Klein -
Cognition, Adaptation and Learning (CAandL) Lab
The CAandL Lab at the University of the Witwatersrand is a research group focused on Computational Neuroscience and Theoretical Machine Learning. The group’s major focus is on understanding child development and how it can inspire artificial intelligence systems. Specifically, the lab focuses on understanding how the structure of the human brain supports efficient learning of language and abstract reasoning. We then aim to develop machine learning models with the same ability for flexible reasoning and generalisation. A major set of tools we employ are those from theoretical machine learning which provide closed-form descriptions of the learning dynamics of artificial neural networks.
Director: Dr Devon Jarvis -
WITS ExplainableAI Lab
As artificial intelligence (AI) systems become increasingly intricate, their decision-making processes, though remarkably accurate, often remain opaque and are often described as "black boxes."This black box nature can prevent users from trusting or using them for problems where there are strong ethical considerations, such as in healthcare, finance, education, and public policy. In these applications, AI's decisions can profoundly impact individual lives and societal trajectories.
Without a clear understanding of how AI systems reach their conclusions, there is a risk that they will be underutilized or, worse, misapplied.
The WITS ExplainableAI Lab (EAI) endeavours to make the complex workings of AI systems transparent, interpretable, and trustworthy. The EAI lab has two main objectives.
- The first objective is to advance fundamental explainable modelling approaches. This is to provide a robust framework for understanding the underlying relationships and dependencies in data, offering clear insights into the 'how' and 'why' of AI decisions, and producing systems that "know when they don’t know."
- The second objective is to transform "black-box" models into transparent systems through post-hoc explainability, making the inner workings and decisions of these black-box models comprehensible and interpretable.
Director: Professor Ritesh Ajoodha
