14 Oct 2019

Go Tsamaya Ke Go Bona: Deep Learning Indaba #3 - A journey just beginning.

Go Tsamaya Ke Go Bona: Deep Learning Indaba #3 - A journey just beginning. By Vukosi Marivate

What is the Deep Learning Indaba?

The Deep Learning Indaba was founded in 2017 with the mission of strengthening African Machine Learning (ML). It was formed by a group of researchers and innovators interested in shaping the future of African Machine Learning.

The vision of the Deep Learning Indaba is of Africans becoming:

  • critical contributors,
  • owners, and
  • shapers of the coming advances in artificial intelligence and machine learning. The organisers come from diverse backgrounds around the continent.

The journey so far

In pursuit of carrying out this mission, the indaba has grown in leaps and bounds in three years.

  • The first indaba was held in Johannesburg at the University of the Witwatersrand, Attracting 330 delegates to kick-start the movement. It is important to note that the African continent had not had a strong, connected community in Machine Learning, Artificial Intelligence, and Data Science. There were regional and even country-specific small communities, but the Deep Learning Indaba’s ambition was to unite the continent and innovate how we all see African Machine Learning. To this end, 22 African countries were represented at the first indaba.
  • The second Indaba had an impressive list of speakers and was held in 2018 at the Stellenbosch University in the Western Cape. The number of delegates increased to 650 – a feat that would not have been possible without the generous cooperation of multiple sponsors as well as a growing hunger on the continent for a movement by us and for us. This Indaba saw the introduction of parallel sessions, which allowed the community to discuss and debate on various themes and to form collaborations.
  • With the success of DL Indaba 1 and 2, it was time to leave South Africa, and Southern Africa, to head to the east and the next indaba was held in Nairobi, Kenya. The journey continued with the announcement of the first Google AI lab on the African continent (in Accra, Ghana) and Microsoft also setting up their first engineering presence on the continent (in Lagos, Nigeria, and Nairobi, Kenya).

Deep Learning Indaba 3, 2019

Indaba 3 was a journey for all involved, with the expansion of the organising committee, a stronger parallel session schedule, a day dedicated to showcasing research (growing from the feedback from past indaba poster sessions) and meeting with the Kenyan ML community. Sessions such as Natural Language Processing, AI Fairness, Reinforcement Learning, Data Science, etc. showcased the breadth and depth of researchers and talent.

As our parents say: Go Tsamaya Ke Go Bona (to travel, is to see in Setswana).

The indaba introduced many of the participants in Nairobi as well as sponsors and allies to the energetic and driven African Machine Learning community. It is one thing to hear about the Indaba, and another to experience it. The Indaba has worked to showcase African ML and drive innovation in teaching and research in the community itself and researchers work till late at night discussing ideas, techniques and sharing their progress and challenges. The Indaba works to serve the community by creating a space for attendees to see themselves as world-class, contributing and shaping AI/ML as emerging technologies that they can own. In many ways, the youth of Africa is moving faster in AI/ML than governments, universities or private industries. As such, the Indaba should also have an outlet to help to mentor the emerging talent.

AI in Africa is self-organising

With the introduction of the Deep Learning Indaba 1 came the introduction of the IndabaX system a few months later. IndabaX’s are smaller, country-specific workshops that run in the first few months of every year. The IndabaX initiative funds smaller workshops to stimulate local communities to gather yearly, to showcase their research and also learn. There are now IndabaX’s in 27 African countries (a great feat for a self-organising programme) – an innovation that has ultimately led to more people learning about the Deep Learning Indaba and allowing local researchers to network and present their research work in a poster session so as to get valuable feedback on their work. The IndabaX also provides a myriad of information about the opportunities and challenges of the ML community across the continent. When the Indaba organisation gets asked about the ML community in a specific country, they tend to connect people with the local IndabaX community. For the IndabaX to be sustainable, one also has to think about connecting local IndabaX communities with their local governments, industry and academic institutions. The Deep Learning Indaba acts as a beacon that we can self-organise and we work to strengthen ML for both research and practice across the continent. Now is the time that local allies come on board to assist in strengthening the local communities.

We are the ones we have been waiting for

We cannot wait; we will not wait! The needs of the African ML community are apparent. Waiting for the “right” conditions, waiting for the right incentive schemes, waiting for “internationally endorsed” approaches will not do. There is no one coming; we are the ones we have been waiting for! The Indaba was not created so that others can come to define African ML. The Indaba was created to let the African ML community speak for itself. The organisers and founders are not those to watch. Those to watch are those attending and those trying to attend the Indaba. The research and innovations are relevant, they are inspiring, and their time has come. We need to strengthen them and help the community chart its own path. Many opportunities are available. We may not have access to research funding in the West, but we can still define how these emerging technologies are introduced and shape our local communities. We may not have enough graduate-level programmes, but we can find ways to empower ML/DS practitioners across the continent to build and showcase their skills. We may not have data protection laws, but we can push our governments to better walk the line between encouraging local data-driven innovation and taking local values into account. We are the ones we have been waiting for!

What comes next?

Deep Learning Indaba #4 2019 will take place in Tunisia. A new adventure awaits! Is the Indaba going to innovate on its programme? YES! Every year is an experiment. The Indaba has already made an impact on the continent and beyond. In Latin America, an Indaba like event KHIPHU has been launched. The same with South East Asia. Indaba Abantu has been contributing and partnering with other organisations like Data Science Africa, Black in AI, Eastern European Machine Learning Summer School and beyond. The Indaba looks forward to many more years of strengthening African Machine learning.

A call to action

As an individual:

  1. You can keep an eye out for Indaba announcements. The Indaba uses the Machine Learning Data Science Africa (MLDSA) mailing list extensively (see link at the bottom of http://mldsafrica.co.za/). 2.Bring others into the conversation and connect potential mentees to potential mentors.

As an organisation:

  1. Encourage and assist your organisation ML/AI/DS members to apply and assist with funding for them to attend Indaba or IndabaX events.
  2. Assist the main Indaba organisation. You can do this by reaching out to the Deep Learning Indaba at https://www.deeplearningindaba.com.
  3. Advocate for a more connected ML/AI community in your country, organisation, etc.

You can also look at some recent coverage on the Indaba:

  • An open letter to the Deep Learning Indaba team – 2019 edition URL
  • Africa Is Building an AI Industry That Doesn’t Look Like Silicon Valley URL
  • Citizen Science URL

Dr Vukosi Marivate is a founder of the Deep Learning Indaba and the UP ABSA Chair of Data Science in the Faculty of Engineering, Built Environment and Information Technology.