Marc Deisenroth travelled to Rwanda to teach statistics at the African Institute for Mathematical Sciences.
In the second installment of a two-part series, he talks about the masters course in Machine Intelligence that he taught, his experiences in teaching it, and his reflections more generally as his four-month sabbatical comes to an end.
During my time at AIMS, I taught a course on Foundations of Machine Learning, as part of the AMMI degree. The course covered mathematical foundations of machine learning and their application to (fairly basic) machine learning algorithms, such as linear regression or principal component analysis. The objective of this course was to uncover the intrinsic mathematical principles, which are often hidden in machine learning algorithms, to allow students to understand how and why these algorithms work and what their underlying assumptions are.
Students are very engaged in class (lots of questions and remarks), and some of our discussions had a lasting impact on my way of thinking and explaining. What I really appreciate (and did not necessarily expect) is that even after a few months, students are excited and enthusiastic about machine learning. I am also happy with the quality of the students. They have a very good mathematical background, and I believe that most students can be strong PhD candidates at internationally leading research institutions.
Due to the pan-Africanism of AIMS, one challenge that everybody at AIMS faces is language, especially students from francophone countries who may have initial difficulties engaging in discussions. The discussions and the engaging conversations with the students definitely highlighted some shortcomings of my teaching, so that I now understand better where and why students struggle. While teaching every day at AIMS is a taxing full-time commitment, it definitely is an experience I will always remember.
As with most new degrees, the beginnings of the AMMI degree were a bit bumpy. For example, we had some early problems with the internet infrastructure, logistics and missing staff. However, together we managed to mitigate all these issues. By now, the hiccups have been resolved and plans for the future are being made.
After the first promising steps this year, in the next year, the AMMI degree will be offered to a total of 100 students, which means a threefold increase of students. Moreover, PhD and Research Fellow positions in machine learning will be offered at AIMS Rwanda, starting in 2019.
AMMI is only part of a growing machine learning and AI community in Kigali. There are many other contributors to the local environment, such as CMU Africa, University of Rwanda but also incubators, space for innovators (eg, K-Lab) and start-ups. There exist already many activities around machine learning and AI, such as the regular AI Saturdays or the DevFest-2018. In future, the local community needs to grow together even more to exploit the synergies in Kigali and Rwanda. A first step toward this is a joint application from different local groups for a one-day IndabaX workshop in 2019.
My personal goal is to establish informal and formal collaborations between AIMS Rwanda and Imperial College, eg, in terms of joint project supervisions, joint PhD supervisions, joint MSc courses. I strongly believe that a symmetric exchange of researchers and teachers will benefit both sides.
After four months, my time at AIMS Rwanda is coming to an end. I enjoyed my time here immensely, and I take with me many positive experiences. It was a great pleasure to see that there are no cultural, gender or societal boundaries when learning together and striving for excellence. I am also very happy that I could help kick-starting and be part of an exciting and unique MSc degree in Machine Intelligence. I wish we had more degrees like this.
I encourage everyone to get involved, either as a tutor or lecturer. You can also remotely supervise projects. If you are interested, please contact the corresponding AIMS branch. Travel, accommodation and food will be covered. Given that the Royal Statistical Society has very close ties with AIMS and can sometimes provide support, it is worth getting in touch with the RSS as well.
I am very grateful to the Royal Statistical Society for their support during this journey. I also thank Imperial College for freeing me from my normal teaching and administrative duties during my sabbatical. Finally, I want to express my deepest gratitude to AIMS, Quantum Leap Africa and all students at AIMS for making my stay an unforgettable and unique experience.
RSS AIMS support is funded by Taylor and Francis. For more information about this initiative please visit: Volunteering with AIMS