Big Data Africa 23
In early March 2023 the Big Data Africa School was hosted in Cape Town by SARAO and DARA Big Data, with support from IBM Research Africa. This was the 4th Big Data Africa (BDA) event, but the first one to be held since 2019 due to the COVID19 pandemic. This school has been very successful over the years at bringing budding young data scientists together from all over Africa, allowing them to develop key skills alongside experts and form important links across the continent. Previous schools have focussed on South Africa and the 8 African VLBI countries, however this year’s event was open to all of Africa. Applications were received from over 30 countries, more than half of the continent! It should also be noted that the majority of applications were from women; previous BDA events and DARA Big Data hackathons have tended to receive around two thirds of applications from men. The increased interest from women was a very welcome development, which led to the school featuring a majority female attendance (18 out of 25 attendees were women). The 25 participants came from 14 African countries, including Malawi, Algeria, Zimbabwe and Uganda. The event was truly an international affair as the tutors also were from a variety of countries; Ethiopia, Namibia, South Africa, France, Spain, Argentina and Kazakhstan. One of the tutors, Dr Lameck Amugongo, is a former DARA Big Data PhD student who studied Cancer Sciences at the University of Manchester. In addition several of the participants had taken part in previous DARA Big Data or DARA events previously; it was clear to see that they were making great progress along their data science career paths.
The theme for this year’s school was healthcare and biomedical imaging, with projects being submitted from prestigious institutions such as the University of Barcelona, Namibia University of Science & Technology (NUST), University of Basel and the University of Cape Town (UCT). The cloud platform (ilifu) and computational resources were provided by the Inter-University Institute for Data Intensive Astronomy (IDIA), who have played a large part in so many DARA Big Data events.
Participants arrived in Cape Town on Sunday 5th March and gathered for a welcome reception in the evening to kick the week off. Introductions were made over dinner and drinks at the Lagoon Beach Hotel, which would be the venue for the week. Participants were introduced to their group members and tutors. Monday started with a speech from Dr Celia Cintas of IBM Research Africa, who was instrumental in the organisation of this event. Dr Cintas outlined the structure of the school and the expectations for the week ahead and asked participants to stand up and introduce themselves. She was followed by Nathi Ndlovu, who works at South Africa’s Council for Scientific and Industrial Research (CSIR) as a Senior Data Scientist. Nathi gave a detailed introduction to the usage of Python for data science, covering environments, methods, containers and more. Prof Karim Lekadir of the University of Barcelona then spoke; Prof Lekadir is the Director of the Artificial Intelligence in Medicine Lab at the University of Barcelona and is the supervisor for some of the tutors who brought projects to BDA. He is also heavily involved in AfricAI, the African Network for AI in Biomedical Imaging, and will chair next year’s MICCAI conference in Morroco.
In the afternoon the tutors presented the projects that they had devised for their groups. The 5 projects were: malaria as a use case; 3D vertebrae reconstruction from 2D X-ray imaging; explainable AI of medical images using SHAP and LIME; combating data scarcity in medical imaging by using AI-generated breast mammogram images; and the application of deep learning techniques for automation of cardiac disease diagnosis. Dr Girmaw Abebe Tadesse, Principal Research Scientist at the Microsoft AI for Good Research Lab, gave participants an introduction to Machine Learning, covering techniques and key areas such as bias and variance in data. Two speakers from IDIA then presented on the ilifu cloud; Dr Jordan Collier talked about the big data needs of the SKA Telescope catalysing the cloud's development, and Jeremy Smith showed participants how they would be using it throughout the week. To end the day Mark Johnson of SARAO talked about pitch decks and how to use them to successfully present projects to clients, colleagues or funders. This session was repeated and expanded upon throughout the week, ensuring that participants left the school with a rounded set of skills.
Day 2 began with a talk given by Prof Yves Wiaux of Heriot-Watt University, titled ‘Synergies between imaging detection techniques in astronomy and biomedical imaging’. He was followed by Prof Tinashe Mutsvangwa of UCT, who spoke about ‘3D bone reconstruction from 2D X-ray images; a solution for improving access to 3D imaging in a resource-limited setting’. Following this the group work began in earnest! The rest of the week was taken up with project work, with teams presenting to each other on their progress at the end of each day. The week was quite intense as groups only had 3 days to solve the problem of their particular project and put together a presentation detailing their methods and findings. Tutors spent lots of time with their teams patiently guiding them and answering their questions.
On Wednesday there was a celebration to mark International Women’s Day 2023. Female participants lined up for photos and all participants were treated to a cake that celebrated the anniversary of 2 previous milestones; the publication of the SARAO Women in Data Science report in 2021 and the resulting Africa Women in Data Science online event in 2022. To continue this theme Melissa Slaymaker, the Director of Women in Tech South Africa, gave a talk about her network and the difference it has made in pushing STEM participation forward for women.
Friday was presentation day; all teams spoke in turn and gave a thorough overview of their work throughout the week. The panel consisted of the key speakers throughout the week plus DARA Big Data fellow Dr Nikhita Madhanpall. At the end of each presentation the panel had many questions and teams had to justify their findings and provide further context. It was clear that a lot of hard work had been done throughout the week and participants had learned a lot. Prizes were given out at a dinner that evening attended by Chris Austin, the Development Director at the British High Commission in South Africa, who said, ‘It is fantastic to hear what you have been working on through this programme, bringing some clever people together from all over Africa. The UK government has been delighted to support the DARA programme over several years and I hope that this legacy will continue’.
Team OOD Red Stars won first prize for their work on using machine learning models to support improved malaria diagnosis. In second place were team Golden GANS, who generated realistic AI images of breast mammograms in order to create more data to be used in research. Individual prizes were also given; Refiloe Shabe (from Lesotho) was given the best team leadership award, Rancy Chepchirchir (from Kenya) was awarded for showing the most progression during the week, and Emmanuel Hansingo (from Zambia) won for best embodying the ‘spirit of the school’ (Emmanuel was a previous participant in the 2020 DARA Big Data hackathon held in Zambia).
To relax from the stress of the week a full day tour of Cape Town was planned for the Saturday; participants got to see seals, penguins and ostriches as well as the Cape of Good Hope and various beautiful beaches before returning home. The event was a huge success, you can read more about participants impressions of the week here. Dr Lameck Amugongo has previously tutored at the 2019 BDA school and also at several DARA Big Data hackathons. When asked about his experience returning as a tutor this time he said, ‘I enjoy seeing participants grasping difficult concepts within a short period of time and applying data science techniques to solve complex research problems. Big Data Africa provides an excellent platform for participants to learn, network with peers and share ideas. Tutoring or mentoring can be challenging because we need to ensure that all students in the group understand the problem, objectives and expected outcomes. This is not easy given the fact that students have different skill levels and are from diverse backgrounds. Nevertheless, it’s a great joy to see the students develop and become comfortable explaining complex problems. Additionally, it is fulfilling for a mentor to see the great work students are able to complete within a week.’
DARA Big Data and SARAO would like to thank each one of our speakers mentioned above as well as all of our tutors and those working behind the scenes to pull this event together; Dr Celia Cintas, Monushia Zimri, Sydil Kupa, Isaac Sihlangu, Dr Kaisar Kushibar, Catherine Namayega, Xolisile Thusini, Dr Lameck Amugongo and Victor Campello.
Read what participants had to say about their experience at Big Data Africa 23 here.
