Speaker: Yvan Dérick KAPTCHOUANG Abstract This talk addresses the challenge of energy efficiency in Graph Neural Networks (GNNs), whose growing complexity leads to high computational and energy costs. We present a generic methodology for designing frugal Graph Convolutional Networks (GCNs)...

Speaker: Asma Basly (OORB Studio) Abstract Robotics development today is plagued by extreme fragmentation-dozens of simulation tools, multiple versions of ROS, incompatible CAD workflows, and countless dependency conflicts that vary across operating systems and hardware configurations. Research teams spend more time...

Speaker Bio My name is AKAMBA MANI Crescence Catherine, and I hold a Master’s degree in Computer Science, specializing in Data Science, from the University of Yaoundé 1 in Cameroon. My thesis focused on predicting credit risk in peer-to-peer lending...

Speaker Bio Benjamin Ultan Cowley is Professor of Learning in Humans and Machines at the Faculty of Educational Sciences, and a Docent of cognitive science. He defended his PhD in Computer Science at the University of Ulster, Northern Ireland, in...

Speaker Bio Evgenii is a doctoral researcher in the HiPerCog group (since 2023), where he combines machine learning with computational modeling to understand human actions in dynamic environments and how learning shapes them. He holds a bachelor’s in Computer Science...

Speaker Bio Pablo, generally known as Pipa, is a doctoral researcher on the CLIC program. In Chile, he completed his teaching degree and went on to teach high-school students in the fields of technology and physics. He joined the HiPerCog...

Speaker: Louis Wei-Yu Feng, University of Cape Town Abstract Existing Large Language Model (LLM) safety benchmarks remain English-centric, severely limiting evaluations for marginalized populations in the Global South. Despite evidence that 85% of women experience online violence, no benchmark systematically...

Speaker Bio Bill Tanekeu is a young Cameroonian graduate, 22 years old. He earned a scientific baccalaureate in 2019 in his hometown of Manjo, on the Cameroonian littoral, before moving to Yaoundé to continue his university studies in computer science....

Speaker: Chris Emezue Abstract Our digital world is a rich tapestry of ideas, languages, cultures, and knowledge. However, our access to and understanding of these resources is skewed; some gain significant visibility, while others remain underrepresented and obscure (even when...

Speaker: Tom Ringstrom Abstract In this talk, Tom presents Option Kernel Bellman Equations (OKBEs) for a new reward-free Markov Decision Process. Rather than a value function, OKBEs directly construct and optimize a predictive map called a state-time option kernel (STOK)...

Speaker Bio A phD candidate at the relational ML group at the CISPA Helmholtz center for information technology. Previously worked on optimizing multi-agent reinforcement learning using ideas from game theory. currently working on a general framework that enables a wide...

Speaker Bio Everlyn is a PhD student in Natural Language Processing at the University of Cape Town. She specializes in Neural Machine Translation for low-resource languages under Prof. Bruce Bassett’s supervision. Her research focuses on data and model-efficient methods for...

Speaker Bio Chinasa T. Okolo, Ph.D., is the Founder of Technēcultură, a Fellow at The Brookings Institution, and a recent Computer Science Ph.D. graduate from Cornell University. Her research focuses on AI governance and safety for the Global Majority, datafication...

Speaker Bio Dr. Daniel Okoh is a Postdoctoral Research Fellow at the Technical University of Kenya under the DARA (Development in Africa with Radio Astronomy) program. He has worked as researcher with the National Space Research and Development Agency (NASRDA)....

Speaker Bio Patrick McSharry is a Visiting Professor at the Department of Electrical and Computer Engineering, Carnegie Mellon University, Research Fellow at the Kigali Collaborative Research Centre (KCRC) and Strategic Advisor to the World Bank funded African Centre of Excellence...

Speaker Bio Batsi is a Master’s student at the University of Cape Town with interests in curiosity-driven reinforcement learning and meta-reinforcement learning. His research focuses on improving the sample efficiency of reinforcement learning algorithms.See you there!

Abstract Recent advances in multimodal training allow for integration of images and text within a unified model. Given their black-box nature, little is known on the strategies developed by vision-language models (VLMs) to allow efficient communication between the two modalities....

Speaker: Emily Jin Abstract One of the key challenges in graph machine learning is how to effectively encode the topology of a graph into the model at hand. Standard message-passing GNNs are known to struggle with counting certain patterns (e.g.,...

Speaker Bio Trained as a mathematician, I then did my PhD in machine learning and computational neuroscience at the University of Oxford, where I investigated the performance of biologically plausible algorithms in deep learning tasks. Following this, I pursued a...

Speaker Bio Dr. Sabrina Amrouche is the co-founder of AfriClimate AI, a grassroots initiative leveraging AI to address climate challenges in Africa. She also serves as Head of Data Science at ZYTLYN, where she leads the development of advanced time...

Speaker Bio I am a dedicated academic and researcher, currently serving as a Lecturer at Wits University. Throughout my academic journey, I sought to merge two of my greatest passions: video games and computer science. These passions led me to...

Abstract Machine Learning is playing an increasingly important role in biomedical engineering. In this talk I’ll discuss some of the hard medical problems we’re solving with data – focusing on our machine learning workflow & how we go from research...

Abstract Understanding and monitoring wildlife behaviour is crucial in ecology and biomechanics, yet challenging due to the limitations of current methods. To address this issue, we introduce two motion capture system specifically tailored for free-ranging wildlife observation. These systems combine...

Speaker Bio I am a Data Scientist at the Bank of Montreal, where my work primarily focuses on modeling volatility in the equity space. Previously, I completed an MSc in Computer Science at the University of Toronto and an MSc...

Online link: https://uct-za.zoom.us/j/92750361177?pwd=QzNiRzBJRjRITVlwa2k5SVNkVmx5UT09

Novel approaches for understanding the neurocomputational basis of interoception and emotion-cognition interactions SAVE THE

TBA

Speaker: Elizaveta Semenova, ML Researcher, Oxford & Imperial College London Abstract Disease mapping is an important surveillance tool that enables researchers and public health officials to analyse the spatial distribution of a disease, identify its geographical patterns, and plan interventions....

Speaker Bio Bruce Bassett has been a Full Professor of Applied Mathematics at the University of Cape Town since 2008 where his research explores both the theory and applications of AI and statistical models. Bruce was formerly head of Data...

Speaker Bio The MARL research team at InstaDeep works on large-scale multi-agent learning with a focus on algorithmic innovation in cooperative systems for industrial applications. The team regularly contributes to the research community through publications at venues such as NeurIPS...

Speaker Bio I have a Ph.D. in physics and currently work as an AI specialist at Allianz. In my spare time, I enjoy researching to understand how our visual system works, approaching it from first principles.

Speaker: Felix Chalumeau, InstaDeep Research Abstract In this talk, we will introduce the challenges of combinatorial optimization and the motivation to tackle them with Deep Learning and Reinforcement Learning. We will walk through some core breakthroughs that happened through the...

Speaker: Callum Rhys Tilbury, Junior Research Engineer @ InstaDeep Abstract Optimising deep neural networks is a challenging task due to complex training dynamics, high computational requirements, and long training times. To address this difficulty, we propose the framework of Generalisable...

Speaker: Divanisha Patel, PhD Candidate @ Wits | AI Research Engineer @ InstaDeep Abstract 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...

Speaker Bio We are the MARL research team from InstaDeep’s Cape Town office. We focus on the most recent advantages of MARL with a focus on JAX-based algorithms and environments

Speaker: Tswelopele, BSc(Hons) Math UCT | MWR CyberSec Abstract This talk will go through some ideas behind my research proposal for my Masters. The proposal puts forth a contribution towards a systematisation of privacy guarantees for machine-learning model-inference with explicit...

Abstract In this talk, I will introduce the ancestral recombination graph (ARG): a powerful way to encode the ancestry of a species through its DNA. ARGs have enabled us to simulate and study evolution on a massive scale, while also...

Abstract I’ll outline an idea I had during our Betty’s Bay getaway, a “thought experiment” in how we might mathematically model symbolic concepts using ideas from category theory. Tali Beynon 17 January 2024

Abstract Robot soccer, where teams of autonomous agents compete against each other, has long been regarded as a grand challenge in artificial ntelligence. Despite recent successes of learned policies over heuristics and handcrafted rules in other domains, current teams in...

Abstract Deep learning has advanced the field of machine translation immensely. However, these advances have not been fully realised for all South African languages, because they are low-resourced and lack sufficient training data. Additionally, the Nguni languages of South Africa...

Abstract AI has never settled on a widely accepted, or even well-formulated, definition of its primary scientific goal: designing a general intelligence. Instead it consists of siloed subfields studying isolated aspects of intelligence, each of which is important but none...

Abstract While reinforcement learning has achieved recent success in many challenging domains, these methods generally require millions of samples from the environment to learn optimal behaviours, limiting their real-world applicability. A major challenge is thus in designing sample-efficient agents that...

Abstract Measurements which characterise urban neighbourhoods have often been collected using traditional survey techniques. This approach, while able to directly capture upstream determinants of health, are expensive and usually difficult to scale across entire cities. On the other hand, routinely...

In this session Batsi and Ruan will share some aspects of their respective research areas. Though this is aimed at the current cohort of UCT honours students taking the RL module, you are invited to attend.

Abstract Deep neural network training is a highly non-convex optimisation problem with poorly understood properties. We know that a solution can be found by following the negative gradient to walk down the loss landscape, but we have little guarantees that...

Abstract 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...

Presented by InstaDeep and AI Society 19 September 2023

Abstract This talk covers a brief intro to Julia programming language. It then compares it to the other options out there for reinforcement learning (and deep learning in general) in terms of usability and speed. Sasha Abramowitz is a research...

Abstract 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...

Abstract The PAILA project, undertaken during our InstaDeep internship, aims to bolster single-environment Reinforcement Learning (RL) algorithms through cross-environment knowledge sharing. To achieve this, we aimed to use symmetric learning agents (SymLA), a meta-reinforcement learning algorithm introducing backpropagation symmetries that...

Abstract Denoising Diffusion Models are a type of generative modelling which serves backbone of recent advances in image synthesis including Dall-E 2, Midjourney, and Imagen. These models utilise an iterative denoising process during inference to produce high quality samples. In...

Abstract Evolutionary robotics lends itself to exploring novel design paradigms in research to assess the efficacy of those designs relative to known paradigms in the space. Origami is one such paradigm that has been relatively under-explored, and has many potential...

Abstract AI safety is a subject which has often been viewed with skepticism regarding its necessity and plausibility in the AI community. However, as we have progressed towards transformational AI systems the urgency of this research has become apparent.In this...

Abstract Interactive learning systems like self-driving cars, recommender systems, and large language model chatbots are becoming increasingly ubiquitous in everyday life. From a machine learning perspective, the key technical challenge underlying such systems is that rather than simple prediction on...

Abstract In nature, morphological diversity enhances functional diversity, however, there is little swarm (collective) robotics research on the impact of morphological and behavioral (body-brain) diversity that emerges in response to changing environments. This study investigates the impact of increasingly complex...

Abstract I will be presenting an empirical exploration of using machine learning and evolutionary algorithms to automate chemical product design. Our study demonstrates how computational design can be controlled via hyper-parameters to generate solutions with desired features and has important...

Abstract Humans are fundamentally social creatures, we live in families, work in teams and our norms of formed from thousands of years of social interaction. What if, along that https://www.youtube.com/watch?v=t3GR91yjOzY Brandon Gower-Winter is a PhD Candi 31 May 2023

Abstract In this talk I want to explain in as clear a way as possible what the key concepts are in a transformer model, explain key terms, and discuss why the transformer is so effective. Watch Benjamin Sturgeon I am...

Abstract The evaluation of empirical algorithm performances in RL appears a closed topic. However, some (sparse) recent research provides unattended criticisms of key elements of the evaluations which are central to the conclusions of many research papers. This talk discusses...

Abstract I explore the challenges facing production AI for health systems in an African context. Progressively I step through the layers of complexity, one can expect to encounter, providing personal insight for addressing some challenges I have found to be...

Abstract We introduce a generalisation of smooth fictitious play with bounded m-memory strategies. We use this learning algorithm to prove a Folk theorem from learning in repeated potential games. If a payoff profile is supported by an m-memory pure strategy...

Abstract Claude presents his work on selective reincarnation for MARL. Claude Formanek 5 April 2023

Abstract Jeremy give’s an overview of PyTorch and Weights and Biases, emphasising how these are useful for ML in production and in research. Jeremy du Plessis 22 March 2023

Abstract Ruan’s highlights and takeaways of NeurIPS 2022. Ruan de Kock 15 February 2023

Visual cortex is optimised for short timescale prediction using spikes Abstract A key question in systems neuroscience is to understand what principles underly the sensory processing throughout the brain. Why are certain neurons in V1 selectively tuned to orientated bars?...

Towards Lifelong Reinforcement Learning through Logical Skill Composition Abstract Reinforcement learning has achieved recent success in a number of difficult, high-dimensional environments. However, these methods generally require millions of samples from the environment to learn optimal behaviours, limiting their real-world...

Generalisation in ML Abstract In Reinforcement Learning there is often a need for greater sample efficiency when learning an optimal policy, whether due to the complexity of the problem or the difficulty in obtaining data. One family of approaches to...

Offline MARL and how to use WANDB effectivly for ML experiments Abstract Claude gave a talk on his research topic, Offline MARL, and also gave a tutorial on how to use Weights and Biases for ML experiments. SPEAKER Claude Formanek...

Abstract Ruan de Kock presents an overview of generalisation in RL.