Schedule

Talk description

In this talk, I will speak about Building Intelligence For Robotics with a specific focus on African robotics and how this can unlock productivity in various different sector.

---

Speaker bio

I am a researcher and technology enthusiast with broad experience in both hardware and software development over various sectors (from self-driving cars in the UK to para-transit in Kenya) with a primary focus on robotics. I strongly believe that technology holds the key to solving some of the world’s and in particular Africa’s most pressing problems. My interests are therefore not only in advancement of technology through research but also in its subsequent education that ensures that this technology is adopted into communities where it would be do the most good. I am currently a Senior Lecturer and Prinicipal Investigator at the University of Cape Town

Talk description

A fundamental talk that highlights some of the foundational concepts that underpin data science, statistical learning, machine learning, and deep learning.

---

Speaker bio

Professor in Mechanical and Aeronautical Engineering at the University of Pretoria, South Africa. Current President of the South African Association for Theoretical and Applied Mechanics, and Chairman of the International Union of Theoretical and Applied Mechanics (IUTAM) National Committee.

Colloquially known as that Gradient-Only Optimization Guy with a published Springer book Practical Mathematical Optimization, co-authored with the late Prof Jan Snyman. Optimization is my Passion, generative modeling my application, and Nvidia GPUs my calculator.

Talk description

We are increasingly living in a world that is characterised by the merger and fusing of the digital and physical realms. ChatGPTs recent massive adoption attests to this. Exactly what this merged reality entails is not clear yet, but there is agreement that almost all aspects of our lives are being affected, either directly or indirectly. Recently some notable Computing scholars described the digital-first observation, which is formally defined as an ontological reversal of the digital and physical worlds [1, 2]. More simply described it is the observation that the physical world preceded the digital world before, but we now observe that the digital world precedes (and often prescribes) the physical world. Anybody that traveled recently would have noticed that our biometric-enhanced virtual identity now precedes any physical document - a physical passport is only used for verification and not primarily for identification.  Several more examples of this digital-first reversal can be observed daily if we care to take notice, including that some of the key objects necessary for learning such as books, music, and videos, now exist primarily and firstly in the digital world before they are being realized into physical existence, and then only when it is necessary for human interaction. This presentation will firstly explore the characteristics of digital objects that inevitably resulted in the digital-first world, followed by a discussion of what digital-first means for us. Lastly, some implications and guidelines are proposed for digital literacy living in this new digital-first world.

---

Speaker bio

Prof Aurona Gerber is an associate professor in the Department of Computer Science in the Natural Sciences faculty at the University of the Western Cape. She is also a group leader of the Centre for AI (CAIR) Research Group in Applied AI. She has published more than 70 accredited publications with co-authors in the key areas of her research that is in various aspects data science and AI, including disruption, ontology engineering, enterprise architecture, and design science research (DSR). She supervised more than 20 masters and doctoral students to completion of their degrees. She is a NRF C1 rated researcher. She is the current past president of SAICSIT, on the steering committee of AISSAC, the SA chapter of the Association of Information Systems (AIS) and SACAIR, the Southern African Conference for AI Research. She is also president of the AIS SIGODIS, the Special Interest Group in Intelligent IS Systems

Talk description

A data-driven approach that leverages latent parameters to simplify the analysis and interpretation of complex data. We showcase the potential of data-driven machine learning and latent parameters in simplifying complex problems and enhancing interpretability. By leveraging the latent space, we enable a more comprehensive analysis of data.

---

Speaker bio

I work as a lecturer at the University of Pretoria (UP). My research interests are in the field of machine learning and guided waves ultrasonics. I am currently in the final stages of my Doctoral studies in the field of artificial intelligence/deep learning, finite element analysis, ultrasonics and rail monitoring.

Talk description

Pushing AI innovation to develop trustworthy explainable AI, AGI and autonomous Intelligent Agents in a Decentralized World

---

Speaker bio

Dr Jacques Ludik is a Smart Technology Entrepreneur, PhD in AI, Founder of multiple AI companies, Author, International keynote speaker, Global AI Ambassador and award-winning AI Leader with a 25+ years’ experience in AI, Machine Learning & Data Science and its applications around the globe. One of his companies, CSense Systems, was Africa’s first AI company that was sold to a multi-national, General Electric, in 2011. Dr Ludik is the author of ‘Democratizing Artificial Intelligence to Benefit Everyone: Shaping a Better Future in the Smart Technology Era’ and ‘Neural Networks and Psychopathology’. He has also introduced an MTP for humanity and its associated goals to help shape a beneficial human-centric future and founded Sapiens.Network as a decentralized, human-centric, user-controlled AI-driven super platform. He is currently involved in sustainable technology venture capital and building AI-based companies with data rich platforms that leverages AI technologies to unlock business, customer and societal value at scale. Dr Ludik is the Founder & CEO of Cortex Group and Cortex Logic, Vive Teen Wellness, Journey Wellness; and Founder & President of the Machine Intelligence Institute of Africa (MIIA) focused on transforming Africa through AI. He was also previously an Executive at Jumo and General Electric as well as an Academic at Stellenbosch University. Dr Ludik currently also serves on the Advisory Boards of Cirrus, Data Science Nigeria, AICE in Kenya, and Verchool and member of the AI Africa Consortium, and is a Swiss Cognitive Global AI Ambassador.

Talk description

There is far more to IBM than servers and networking, and this talk will cover some of the local projects and ventures that are happening at IBM Research & IBM Quantum. If you are keen to discover what opportunities or potential research collaborations are possible, then join us for our talk.In this part of the talk, Stephanie will touch on the numerous opportunities available at IBM Research, the domains for which we develop technology solutions, including Healthcare, Quantum, AI and Foundation Models. Quantum computing may sound intimidating to some, but, if you have some linear algebra along with a good dose of curiosity in your back pocket, then you are ready to entangle some qubits. Gemma will guide you to resources, events, certifications, and community projects to help you get started in this thrilling field.

---

Speaker bio

Gemma

With degrees in Statistics, Applied Mathematics, and Operations Research, Gemma is a programme manager for two IBM Quantum global programmes (Qiskit Advocates and the Qiskit Developer Certification). The first ten years of her career was spent in education before changing to a career in tech where she worked in a number of different roles. What remained consistent was a passion for community building and that is what brought her to IBM Quantum and the rapidly growing field of quantum computing.

Stephanie

With degrees spanning human biology, bioinformatics, and genetics, Stephanie has always had a passion for finding solutions for healthcare problems plaguing our communities. As the Technical Lead for Healthcare and Life Sciences Research at IBM Research in Johannesburg, Stephanie leads a diverse team of individuals specialising in machine learning, AI, mathematics and quantum chemistry.

Talk description

Education isn’t a transaction, it’s a process driven by creativity, curiosity, and connection. In a world where our children will grow up as AI natives, a new ally emerges in the quest for effective learning and boundless curiosity - artificial intelligence (AI). What does the future hold for our learners, teachers and schools?

---

Speaker bio

Gabi Immelman is the founder and CEO of edtech start-up and social learning platform Mindjoy. A former Reggio-educator turned EdTech founder with experience working globally in Africa & USA. Mindjoy is unleashing curiosity by building an OS for boundless classroom learning powered by AI. We’re starting by helping educators transform lessons they love and know and turn them into tech projects beyond their imaginations. Mindjoy is the third company she has founded - her first being a stick-on tattoo parlour at the beach where we use to go for summer vacation when I was a child.

Talk description

I’ll discuss the forthcoming revolution in radio astronomy powered by new-generation radio telescopes like the South African MeerKAT and the Australian ASKAP facilities. The massively increased data rates make Machine Learning techniques an ideal tool to tackle radio astronomical data. As a prime example, I’ll talk about the problem of astronomical source detection and classification.

---

Speaker bio

Kshitij Thorat is a radio astronomer and has worked as a senior lecturer at the University of Pretoria. He finished his doctoral work at the Indian Institute of Science in 2014, after which he has been based in South Africa. His research interests revolve around extra-galactic radio sources: their life-cycles, morphology and their impact on their surroundings. With the advent of the Big Data paradigm in radio astronomy, Kshitij has focused on techniques to help process and analyse data better, including development of fully automated data reduction pipelines and machine learning techniques to classify radio galaxies.

Talk description

In this in-depth introduction to the Transformer architecture, we will be exploring attention and its crucial role within Transformers. From there, we will dive into the essential aspects necessary to truly understand how a Transformer processes various types of data, going beyond just text. Furthermore, we will touch upon the exciting future developments and possibilities of Transformers. In addition, we’ll take a closer look at how ChatGPT works, providing you with a comprehensive understanding of this current phenomenon.

---

Speaker bio

I am co-founder and technical lead at ByteFuse, and I have an unhealthy obsession with cats and llamas.

Talk description

If you’ve ever taken a statistics class, I have little doubt you’ve heard someone say “correlation does not imply causation”. This trope seems to imply that we can never state that A causes B by analysis of data alone. This talk will take a deeper look at if this is always the case by introducing the field of causal inference, with a special focus on (graphical) causal modelling. Attendees will leave having gained knowledge of how assumptions and data come together to enable drawing causal conclusions from data. More importantly, they will develop the basic intuition and fundamental concepts of causal modelling in the context of machine learning.

---

Speaker bio

St John is incoming PhD candidate at The University of Cape Town, where his research focus is in the area of reinforcement learning and learning agents. His masters thesis investigated the causal methods in the context of reinforcement learning. This led to an interest in how causality can aid in helping solve various problems faced in adaptive learning methods. He also has an academic interest in neuro-plausible methods and multi-agent scenarios, with a strong interest in the emerging field of active inference.

Talk description

GANs have been used for a variety of applications. In this talk I will discuss how GANs can be used for both semi-supervised learning and novelty detection in one model. Both these tasks are becoming increasingly important in research and industry, but few have synthesised GAN models to accomplish all in one. Unified machine learning is critical for the just development of Aritificial Intelligence, and this talk will discuss one such step.

---

Speaker bio

PhD Candidate in Electronic Engineering at Stellenbosch University focussing on Machine Learning. My specific interest is Generative Adversarial Networks which are at the forefrunt of the 4th Industrial Revolution.

Talk description

Discover the fundamentals of contrastive learning, a technique that learns meaningful representations by contrasting similar and dissimilar samples of words, sentences, or images. In this talk, an introduction to both supervised and self-supervised contrastive learning will be provided, as well as an outline of its applications in Computer Vision, NLP, and recommender systems.

---

Speaker bio

I am a Master’s student at Stellenbosch University pursuing a degree in Machine Learning and Computer Vision under Dr Willie Brink.

Talk description

Arthur C. Clarke once said, ‘Any sufficiently advanced technology is indistinguishable from magic’. One could argue that data scientists are modern magicians… but are we really? The role of a data scientist in industry spans from understanding business context and identifying problems to solve, to thinking of meaningful metrics and explaining solutions to stakeholders. If you are a grad wondering what data science in the real world looks like, this talk is for you.

---

Speaker bio

Sedinam is a data scientist in the UK AI and Data team at Investec. She graduated from the University of Cape Town with an honours in statistics and is on her way to study a masters in machine learning at the University of Cambridge.

Talk description

Small and big data engineering lessons in industry.

---

Speaker bio

Data Engineer at Teraflow.ai.

Talk description

As ML models adoption continue to grow in real world applications, model explainability is becoming equally important. It is necessary for data scientist to be able to explain the models in production.

---

Speaker bio

Dineo is currently a Senior Data Scientist at Vodacom Group with an understanding of machine learning and its applications across diverse industries. Driven by her passion for making a positive impact, Dineo has worked in healthcare research, banking and telecoms, leveraging data-driven solutions to drive innovation and tackle complex problems. She’s highly motivated by the prospect of transforming cutting-edge machine learning techniques into practical, scalable solutions that improve efficiency, decision-making and ultimately, the lives of the end-users.

Talk description

Navigating the Intersection of Social Science, Algorithms, and AI-Driven Technologies: Explores the idea that human beings are essentially complex algorithms that have evolved over time. It covers the development of human intelligence and how it has been influenced by the emergence of Artificial Intelligence (AI) technologies. Covering three key areas—social science’s interplay with human technology, the impact of algorithms on humans, and the role of artificial intelligence in automating human intelligence—the discussion aims to address the challenges and opportunities presented by these interactions. By understanding and addressing these concerns, we can work collaboratively to shape a future that prioritizes the well-being of humanity while driving ethical innovation.

---

Speaker bio

Jay van Zyl is a founder at Ecosystem.Ai

Talk description

Mastering Modern Computer Vision: From Basics to Cutting-Edge Techniques’ workshop provides a comprehensive exploration of core and advanced aspects of Computer Vision. The tutorial will delve into essential techniques such as object detection, segmentation, pose estimation, and text detection, all within a deep learning framework. This high-impact session aims to bridge theory with practical application, equipping attendees with the skills needed to leverage Computer Vision in solving complex real-world challenges.

---

Speaker bio

Ritesh Kanjee is a seasoned AI entrepreneur, recognized educator, and the Founder of Augmented Startups, a leading platform for AI and Computer Vision learning.

Talk description

Tasks such as speech recognition and speech synthesis are increasingly being tackled with large spoken language models. So what is left in speech processing research? In this talk I will discuss some of the less conventional tasks that we are tackling with these language models. Our long-term goal is to try and use these models to gain a better understanding of how humans acquire language, and to improve our scientific understanding of the speech signal.

---

Speaker bio

I am an Associate Professor in E&E Engineering at Stellenbosch University. Before this, I did a postdoc at TTI-Chicago. I obtained my PhD in 2017 from the University of Edinburgh. I am a researcher in machine learning with a specific focus on speech processing. My goal is to develop methods that would allow machines to understand speech without explicit supervision. Through this, I hope to better our understanding of human language acquisition and what is going on in the speech signal itself.

Talk description

Voice conversion aims to transform speech into a target voice with just a few example recordings of the target speaker. Recent methods produce convincing conversions, but at the cost of increased complexity – making results difficult to reproduce and build on. Instead, in this talk I will go over some of our lab’s recent research to keep voice conversion simple by using just the plain k-nearest neighbours algorithm. Despite its simplicity, our recent research is competitive with the largest and complex existing models. I’ll also show how it can be used to perform diverse combinations of conversions, such as converting cross-lingually, or between human and animal sounds.

---

Speaker bio

Matthew is an Electronic Engineering PhD student in the Stellenbosch University MediaLab under Herman Kamper. His research focus is on speech synthesis and generative speech models.

Talk description

We propose a visually grounded speech model that learns new words and their visual depictions from just a few word- image example pairs. Given a set of test images and a spoken query, we ask the model which image depicts the query word. Previous work has simplified this few-shot learning problem by either using an artificial setting with digit word-image pairs or by using a large number of examples per class. Moreover, all previous studies were performed using English speech-image data. We propose an approach that can work on natural word-image pairs but with less examples, i.e. fewer shots, and then illustrate how this approach can be applied for multimodal few-shot learning in a real low-resource language, Yoruba. Our approach involves using the given word-image example pairs to mine new unsupervised word-image training pairs from large collections of unlabelled speech and images. Additionally, we use a word-to-image attention mechanism to determine word-image similarity. With this new model, we achieve better performance with fewer shots than previous approaches on an existing English benchmark. Many of this model’s mistakes are due to relating a few-shot class with its contextual information. We apply multimodal few-shot learning, for the first time, on an actual low-resource language, Yoruba. By initialising the multimodal Yoruba few-shot model on a larger set of English speech-image data, it can outperform its purely English variant.

---

Speaker bio

I am currently a DeepMind scholar and a PhD student in Electrical and Electronic Engineering at Stellenbosch University. My thesis investigates the usage of vision as a form of transcribing speech in a low-resource language. With this research we hope to also get some cognitive insights as to how children learn.

Talk description

In this talk I will present some interesting AI use cases in the BMW Group South Africa and also discuss the transformative power of MLOps and Low Code No Code.

---

Speaker bio

I am currently employed as a data scientist at the BMW Group South Africa. Before this, I was in academia in the Department of Statistics, University of Pretoria.

Talk description

I’ll discuss how machine learning solutions are implemented in enterprises, what the role of a machine learning engineer is, and how they leverage MLOps to be successful.

---

Speaker bio

After finishing my PhD in Operations Research, I worked on projects in superconducting electronics and extremely high-frequency data capture. In 2014 I helped found the precursor to what is now Spatialedge. Currently I serve as Director and COO at Spatialedge and help various teams deliver high-value MLOps projects at various large companies.

Talk description

An introduction to the way that the technologies powering ChatGPT are also impacting biology and medical research. By using Ai sequence models - we can now model protein interactions, predict vaccine candidates, and even generate new medicines.

---

Speaker bio

I studied NLP at UCT, and now I work for Instadeep, under BioNTech - on the development of Ai tools for Vaccine R&D.

Talk description

What does it mean to speak of personal identity, autonomy, and agency in the light of the current status and exponential growth of data-driven AI technologies? Why does the question matter?

---

Speaker bio

Professor Emma Ruttkamp-Bloem is a philosopher of science and technology, an AI ethics policy adviser and a machine ethics researcher.

Talk description

Bayesian inference allows us to calculate the posterior distribution of unknown variables given observations, using Bayes’ Theorem. In practice however, it is typically the case that this posterior distribution is intractable to compute exactly. This tutorial introduces variational inference (VI) - which constructs an approximate posterior by framing the inference problem as an optimization problem. As part of this tutorial, we will consider tools such as Pyro which help simplify the application of VI and allow it to scale to large datasets and models. The tutorial concludes by considering an example application of VI in deep learning - Variational Autoencoders.

---

Speaker bio

MSc graduate from Stellenbosch University, with a thesis focused on the integration of graphical knowledge in the form of Bayesian Networks into deep learning models such as Normalizing Flows and Variational Autoencoders. Currently working at Capitec Bank as a machine learning engineer.

Talk description

Introduction to the Centre for High-Performance Computing and its capabilities to accelerate Deep Learning Projects

---

Speaker bio

The Centre for High Performance Computing (CHPC) is a national high performing computing facility that is funded by the Department of Science and Innovation (DSI) and administered by the CSIR.

Talk description

In this talk we will demystify how machine learning has been used to improve the efficiency and also do planning for solar energy. We will review the past and propose possible avenues for the application of machine learning in solar energy.

---

Speaker bio

I am a research masters student at Stellenbosch University (Deep-mind Scholar) in the department of applied mathematics. I hold a masters in mathematical sciences from the University of the Western Cape. I did a BsC in Mathematics and Computer science in Kenyatta university, Kenya. My interests rests at the intersection of mathematics and computer science. I love working towards solving a real life problems such as in solar energy generation efficiency.

Talk description

The talk highlights the importance of developing text-to-speech (TTS) technologies for African languages, particularly in resource-scarce environments. It explores the challenges and potential solutions for addressing the linguistic diversity of African languages through TTS systems, ultimately aiming to bridge the digital divide and promote inclusivity in the region.

---

Speaker bio

I completed my studies in electronic engineering at the University of Pretoria and immediately ventured into the field of speech technologies, specialising in text-to-speech (TTS). With over 15 years of expertise as an accomplished C developer, I have successfully executed the development of portable medium to large-scale projects. My proficiency extends to interfacing low-level C libraries with both interpreted and compiled languages. I possess hands-on experience in implementing various machine learning techniques, ranging from traditional pattern recognition to cutting-edge deep learning, in both research and commercial systems. Additionally, I have extensive knowledge in the application of digital signal processing within commercial-grade software. Throughout my career, I have effectively led small, diverse teams consisting of highly skilled researchers and developers, driving them towards shared objectives. I am the principal architect and engineer behind the Qfrency text-to-speech system, which marks a significant milestone as the first commercial-grade TTS engine capable of synthesising text in all 11 official South African languages. The successful commercialisation of the Qfrency TTS system has facilitated its widespread adoption across various sectors, both domestically and internationally. Notable fields of implementation include education, assistive technologies, call centres, and more.

Talk description

In the intersection of Software-Defined Networking (SDN) and Machine Learning (ML), we’ve embraced a new paradigm for smart network management. SDN, with its unique approach of decoupling the control and data planes, enables a centrally managed system that streamlines traffic flow. But the true breakthrough emerges when we integrate this with ML. This fusion empowers our system to classify encrypted network traffic, master the complexity of expansive networks, and make intelligent, self-regulating decisions. In essence, our framework is redefining network management by combining SDN and ML, delivering improved user experience through intelligent, autonomous networks.

---

Speaker bio

Keegan White has a Masters in Computer Science for UCT. He is the founder of a start-up that focuses on integrating machine learning with computer networks through the software defined networking paradigm. He is also a director of iNethi, a non-profit that aims to empower local communities to create, maintain and expand Wireless Community Networks. His expertise lie in deep learning and computer networking.

Talk description

An overview of how Peralex Electronics employs Deep Learning in radar, communications and acoustic signal processing applications.

---

Speaker bio

Thomas is a Radar and Machine Learning Engineer at Peralex Electronics. He recently completed his MSc degree in Radar and Electronic Defence at the University of Cape Town, where his research focused on radar imaging. In his role, Thomas utilizes Deep Learning in radar, communications, and acoustic systems across a wide range of applications. He is particularly interested in the intersection of Engineering and Artificial Intelligence, aiming to develop advanced technologies that can make a positive impact on society.

Talk description

Liveness detection is a safeguard which plays a crucial role in the financial services industry for biometric fraud prevention. This talk will provide an introduction with a particular focus on the two main variants of face liveness detection, namely, active and passive liveness detection. Real-world scenarios, including deepfakes and other types of liveness spoofs will be discussed.

---

Speaker bio

During his time completing a B.BusSci degree, specialising in Analytics at the University of Cape Town, Ali developed a keen interest in machine learning and a strong passion for NLP. Whilst working as a machine learning engineer at Sybrin, his role has encompassed a vast number of topics related to computer vision and NLP. However, his primary focus has been biometrics research and development. As an ambassador for DeepLearning.AI, he also hosts occasional community events to help further the AI scene in South Africa.

Talk description

An explanation of the fundamentals of reinforcement learning, touching on multi-agent and multi-objective reinforcement learning

---

Speaker bio

Asad Jeewa is a lecturer of Computer Science in the School of Mathematics, Statistics and Computer Science at the University of KwaZulu-Natal. He previously served as a researcher at InstaDeep and IBM Research Labs. He is currently pursuing a PhD at UKZN in Reinforcement Learning. Asad holds industry certifications in Software Development and Deep Learning. He supervises MSc and Honours students and reviews for international conferences.

Talk description

Convolutional Neural Networks are at the heart of state-of-the-art detection, recognition and tracking of objects in images and image sequences, surpassing the capabilities of traditional neural networks in this domain by far. This talk takes a very visual and unconvoluted approach to describing how CNNs work from the ground up, describing the fundamentals and key concepts, and building up to describing some established CNN architectures that have proven their mettle in a range of vision-based tasks. The talk ultimately aims to provide a good intuition about how CNNs work and demystify and deconvolute key CNN terminology.

---

Speaker bio

Dr Mehrdad Ghaziasgar is a Senior Lecturer in Computer Science at the University of the Western Cape. He has more than a decade of experience in research in the domain of applied machine learning, with over 40 accredited publications and an extensive postgraduate supervision track record. His research group has produced over 45 assistive technology prototypes, some of which have received national acclaim. He also consults to industry in the same capacity. He is particularly enthusiastic about finding ways to convey complex topics in Computing to diverse audiences.

Talk description

Driving Innovation, Empowering Communities: Capacity Building for Inclusive AI in Africa

---

Speaker bio

John Kamara is a Tech Entrepreneur with over 20 years’ experience working in new market acquisitions across various technology verticals in Europe, America, Asia and Africa.

John is one of Africa’s leading experts on how businesses can leverage key technology trends, transform organizations and drive competitive advantage for impact in industries including finance, agriculture, health, education, gaming and startup enterprises.

John is actively involved and heavily invested in new technologies focusing on IoT (Internet of Things), AI (Artificial Intelligence), Blockchain and Crowdfunding. Having worked in market growth leadership roles for global technology companies at Google, Conduit, eir Ireland, Digicel, BMC Software and others, John is committed to seeing technology play a critical role catapulting Africa into the Fourth Industrial Revolution while creating critical mass impact.

Talk description

Denoising Diffusion Models are a type of generative modelling which serves backbone of recent advances in image synthesis including Dall-E 2 and Imagen. These models utilise an iterative denoising process during inference to produce high quality samples. In this talk we explore the fundamentals of diffusion models, the intuition behind them, how they work in practice, and how they may be generalised to a wide range of applications.

---

Speaker bio

MEng graduate from Stellenbosch Univeristy, with a thesis focused on combining speech processing and multi-agent reinforcement learning. Currently working at InstaDeep as a BioAI research engineer.

Talk description

This talk will discuss what makes Julia so applicable to RL, how it could benefit your future projects, what it’s competitors are and why it’s not yet as widely used as python.

---

Speaker bio

Sasha is an MSc graduate from UCT, with a thesis focused on evolutionary methods for reinforcement learning. He is currently a research engineer at InstaDeep on the multi-agent team.

Talk description

In this talk we focus on methods to instil curiosity in reinforcement learning (RL) agents, as a form of intrinsic motivation. We will start by exploring some of the challenges in RL, such as dealing with sparse rewards. To overcome these challenges, researchers investigated adding intrinsic motivation to agents. After defining intrinsic motivation, and its impact within RL, we’ll delve into the pivotal role curiosity plays. We will trace the evolution of curiosity algorithms and learn how they act as a driving force for intrinsic motivation. Lastly, we’ll showcase cutting-edge research and methodologies that harness curiosity to enhance RL agents’ performance. From overcoming challenges to unlocking new possibilities, curiosity opens the door to transformative advancements in RL.

---

Speaker bio

Batsi Ziki is a research intern at InstaDeep and concurrently pursuing a Master of Science degree at the University of Cape Town, under the supervision of Professor Jonathan Shock. Batsi’s current research interests revolve around Meta-Reinforcement Learning, a field that explores the development of intelligent agents capable of adapting to any new task with as little data as possible. Beyond his academic pursuits, Batsi enjoys immersing himself in the world of video games, indulging in captivating films, and taking walks.

Talk description

This talk will provide an introductory overview of reinforcement learning and its key concepts. We will then focus on how InstaDeep is using reinforcement learning to solve real-world problems.

---

Speaker bio

I am a PhD student at the RAIL Lab (University of the Witwatersrand) and an AI research engineer at InstaDeep. My PhD research focuses on model-based reinforcement learning - more specifically, I am looking at how models for intuitive physical reasoning tasks can be learned and then used within a reinforcement learning framework. My work at InstaDeep involves applying reinforcement learning to solve the combinatorial optimisation problem of 3D bin packing.

Talk description

In the rapidly evolving landscape of deep learning technologies, it is imperative to emphasize the importance of human rights and the principle of doing no harm. While ethics serves as a crucial foundation, it is not enough to ensure the protection of individual rights and dignity. Human rights must be the guiding principle throughout the development and deployment of these powerful algorithms.

---

Speaker bio

Dr Eileen I Carter is an accomplished legal professional, serving as the Provincial Manager of the South African Human Rights Commission in the Eastern Cape. As an admitted attorney and accredited mediator, she has dedicated her career to promoting access to justice for marginalized communities. She boasts with several academic, professional and leadership accolades and has presented on local regional and international platforms on issues relating to the rights of vulnerable groups. She is currently pursuing studies on the intersection between human rights and developing technologies.

Talk description

My talk will offer an introduction to the legal and licensing landscape for machine learning projects and application and will engage participants on the legal and licensing challenges faced in their work, proposing solutions and getting feedback/input as to the workability of proposed solutions.

---

Speaker bio

Chijioke Okorie founded and leads the Data Science Law Lab - a research group at the University of Pretoria’s Centre for Intellectual Property Law that deploys research in law and produces independent and authoritative evidence and policy advice to support the growth of all aspects of data science research, development and application across Africa. Before becoming a full time researcher and policy consultant, Dr Okorie was a Senior Associate in the law firm of CLP Legal, Lagos Nigeria. She is also the Lead Advisor at Penguide Advisory, a firm that provides research, educational, legal and regulatory advisory services in various fields including copyright and (emerging) technologies. At Penguide, she has the responsibility of leading research, training and legal advisory services for a range of clients including government and government agencies, companies and non-profit organisations, entrepreneurs within Nigeria, across Africa and internationally. Dr Okorie is ‘Africa Correspondent’ at award-winning intellectual property law blog, The IPKat and an associate editor of the South African Intellectual Property Law Journal. More information about Dr Okorie can be found at www.chijiokeo.com

Talk description

Enabling your People to do their best everyday, is so often forgotten in the frantic pace of Tech start up’s. I will share valuable lessons learnt over a successful global career on how to create an ideal environment for business and employee growth and success.

---

Speaker bio

Roger’s leadership track record spans, South Africa, the Middle East, Europe and most recently the US. He has partnered with CEO’s Founders, Boards of Directors and Investors across a wide variety of business sectors in local, regional and global roles. His passion is to shape strategy, unlock potential , build teams and deliver success . He recently returned to Cape Town and is now the Chief Human Resource Officer for DataProphet, an award winning Cape Town based Artificial Intelligence for Manufacturing Scale Up. Roger is a certified Executive Coach , and lives in Paarl where he enjoys the beauty of the Cape Winelands.

Talk description

We are often prisoners of our own assumptions and let others take control of our life. The talk is about some important decisions to take, especially as women, to be in control of our life, career, future. It is an honest reflection of the choices we have to make and how to best prepare.

---

Speaker bio

Andrea Böhmert is a partner at Knife Capital, a Venture Capital company based in Cape Town. By leveraging knowledge, networks & funding, they focus on supporting the international expansion of South Africa’s high growth scale-ups. Having grown up in Germany, Andrea came to South Africa in 1994. The early years of her professional career were spent in Corporate SA, as Head of Strategy for Siemens Southern Africa and Manager of a Software Division at Dimension Data. In 2007, she entered the Venture Capital space by single-handedly raising a R350m VC Fund, Hasso Plattner Ventures Africa, from the co-founder of SAP, Hasso Plattner. In 2011, she joined Knife Capital as Partner. Since then, Andrea has been instrumental in many eco-system defining initiatives in South Africa – representing South Africa’s VC system on numerous local and international panels. She has recently been appointed to the Board of SAVCA, the South African Venture Capital Association. Her goal is to prove that South African founders and the companies they build can compete on a global level. She sees herself as a business builder and a business partner, captivated by founders who have a deeper motivation and embrace the journey. Andrea knows that one needs an ecosystem to succeed, and is trying to play her part in building that.

Talk description

We as ladies are a rare breed within this beautiful and futuristic sector. Perhaps, with some introspection, we can do even more.
I’ll address some developmental opportunities I’ve noticed, such as imposter syndrome, lack of female mentors and lack of networking sessions.

Talk description

This presentation focuses on being intentional about personal development and career growth, particularly in the field of Artificial Intelligence (AI), with a focus on societal impact. The talk is aimed at inspiring individuals to consider their roles in shaping the future of AI and its implications on society.

---

Speaker bio

Mmaki Jantjies holds a PhD in computer science with research interests broadly in the field of technology for development and is an NRF rated researcher.

She is the Group Executive for Innovation at Telkom and is also an adjunct associate professor at the University of the Western Cape.

Talk description

This talk is on AI Alignment, Safety and Preparedness progress in Africa. It will look at cutting-edge alignment and FAIR (Fairness, Accountability, Interpretability, and Robustness) progress in the context of AI safety. We are excited to explore the potential for African countries to lead in responsible AI development and safety. In this session, we will delve into key aspects, such as community engagement, public awareness, and initiatives in low-resource languages. Offer recommendations for strengthening AI safety research, education, policy frameworks, and collaborations. Discuss the potential for African countries to become leaders in responsible AI development and safety.

---

Speaker bio

I am Jonas, the founder of the African AI Safety Lab, Equiano Institute, and Progress Studies Journal. I am passionate about promoting AI alignment and policy for African.

I pursued my studies in Computer Science, earning a candidate for Masters at the University of Sussex and completing a BSc in Mathematics at Istanbul University. As an independent researcher, I established the Progress Studies Journal, which serves as an overlay journal dedicated to distilling research aimed at technological progress, and responsible innovation.

Previously, I served as Founder in Residence at Enterpreneur UK in London, where I worked with a select group of entrepreneurs to build projects and foster partnerships. I am also actively involved with Cohere For AI as a volunteer, leading AI Safety Talks. In this role, under Sara Hooker.

Talk description

An introductory discussion on AI governance frameworks on the African continent and their status alongside global efforts in mitigating and preventing AI risks. The aim of the presentation is to make a case for the urgent need for governing AI by our governments and international regional community bodies such as the AU, SADC, COMESA, ECOWAS and AMU.

---

Speaker bio

Zainab is a final year LLB student at the University of Cape Town and a graduate of Rhodes University having majored in Geography and legal theory. She is the current chairperson of the Effective Altruism student society at the University of Cape Town. Zainab is interested in issues of governance the development of effective policy for positive impact. She is currently involved in researching current proposals and drafting of global AI governance frameworks.

Talk description

This talk discusses the key arguments for why powerful AI systems may pose catastrophic risks.

---

Speaker bio

Ben, based in Cape Town, used to work as a machine learning engineer. He’s now focused on studying AI safety, using a research grant from the Long Term Future Fund. His research currently focuses on trying to measure levels of agency in AI agents in RL settings. He’s also interested in large language models and how they might be used to influence the behaviour of RL agents.

Talk description

A Data-Driven AI Classification Model for Health Care Applications.

---

Speaker bio

Ayogeboh Epizitone is a researcher and a PhD candidate whose research goals and achievements have been centred on Information Technology in the 21st century. She has multiple certificates, diplomas and a master degree. Her experience and background cover IT, Accounting, Finance, Business, and Management sciences. She is also a Durban University of Technology’s ICT and Society Research group member. She holds a special inclination towards scientific computational methods in the specialized areas of machine intelligence, computer systems, software implementations and data sciences. And she is dedicated to incorporating scientific cutting-edge data-intensive techniques to generate pertinent, astute and valuable solutions to society’s acute problems. Hence her current and future work and interests from this angle delve into the use of data science, Machine learning and artificial intelligence with an emphasis on building long-term intelligent, resilient and sustainable systems for efficient and effective applications.

Talk description

The biomedical domain has a plethora of unstructured data in different forms (clinical notes, image scans, laboratory data, and electrical impulse scans). We present an overview of biomedical text insights extraction for structuring clinical notes.

---

Speaker bio

Siphiwe Thwala is a Research Scientist within the computational linguistics team at IBM Research Africa, South African Lab. His research background is in computational astrophysics, with a focus on using unsupervised learning to detect signals of the plasma produced by non-thermal diffuse emission, in extended radio sources. His work at IBM includes: semantic based sentiment analysis on twitter data; interactive data visualisation of twitter data; and semantic based insights extraction and knowledge modelling on biomedical text.

Talk description

The effectiveness of machine learning models relies heavily on the quality of the data they are trained on. However, managing data quality presents well-known challenges, including manual, time-consuming, and resource-intensive processes. In this presentation, we will explore how machine learning techniques can be utilized to enhance data quality.

---

Speaker bio

Junior Muka is a passionate data management professional with over 15 years of industry experience. He possesses deep expertise in data governance, data quality management, and big data architecture. He heads up data governance at Woolworths and regularly shares his experience with students as a part-time Lecturer at the University of Western Cape (UWC). Junior is a PhD candidate in Computer and Information Sciences at the University of Arkansas at Little Rock (UALR), USA.

Talk description

The fields of customizable pinpoint medicine and healthcare research have recently undergone a revolution thanks to artificial intelligence and machine learning. Large amounts of health data are now available, including patient medical records made accessible by wearable sensors, health insurance claims, medical imaging, surveillance, and more. The rapid development of machine learning algorithms, which primarily use deep learning techniques to analyze clinical data, is gradually improving disease surveillance, enabling doctors to make better diagnoses, facilitating early disease detection, uncovering novel treatments and drug interactions, and other effects. Research and development in this area are moving more quickly than before.

---

Speaker bio

Dr Julius Oluwasegun Aroba, a Durban University of Technology lecturer, Department of Information Systems. Honorable Research Associate in the Department of Operations and Quality Management. He has a PhD. in Information Technology. A graduate of Information Technology University from the prestigious Coventry University United Kingdom, BSc. Computer Science and Technology (Upper-Class Division) Crawford University whose research inclination focus area are into Wireless Sensor Networks, Hyper-Heuristic, Hybrid-Heuristic, Meta-Analysis, Project Management, Data Science, Machine Learning, SAP Specialist with over decade years of experience in the higher education sector, healthcare industries, government parastatals, a consultant across the globe, a graduate member of IEEE, IITPSA South Africa, Member of IET UK. He has Chaired and Co-Chaired International and National Conferences. He is a global mentor to SMEs and European African, Sisonke NdabaX, and a seasoned guest speaker. He has a vast collaboration network across Spain, France, Morocco, the USA, the UK, South Africa, and Sweden.

Talk description

The talk discusses the problems in applying deep learning to predict CVD multimorbidity in Africa. Accuracy issues and explainability limitations will be explored, which will be followed by proposed solutions using emerging techniques and algorithms.

---

Speaker bio

Dr Kayode Adetunji holds a PhD in Electrical and Information Engineering from Wits university, worked on computational and artificial intelligence algorithms. He is currently a Postdoctoral research fellow at the Sydney Brenner Institute for Molecular Bioscience (SBIMB), working with the MADIVA team. His research focal point is to develop advanced machine learning models for understanding the patterns of multimorbidity using phenotypes, genomic and other etiological characteristics.