Data Science Nigeria 2019 Microsoft Deep Learning Bootcamp: My Experience.



I’d been in the Data Science Space for about a year now. It had been a rough road, for a young man coming from a Pharmacy background, with little knowledge of Maths and Statistics. AI never forgives, AI never forgives.

First, like always, i’d say thank you to Dr Bayo Adekanmbi for his drive and commitment to this program. I wish i had a better way to express my gratitude, i really wish i did. Thank you Mrs Toyin Adekanmbi as well for overseeing our welfare(the meals were super delicious!!!) while the bootcamp was going on. I saw the passion with which she made sure things went according to plan, i’m not taking this lightly, thank you ma. Lastly, my gratitude goes all of our instructors who flew in from the UK, US and South Africa to come equip us with AI skills.

My story starts here. So the bootcamp kicked off on Wednesday(10th of October) with Mr Karim Beguir, CEO Instadeep, who flew in from London. Karim made a great presentation about how machines can now go beyond human intuition because of increased amounts of data, better computational power and sophisticated algorithms. I heard about ‘AI at the edge’ for the first time in Karim’s class and i’m looking forward to using it to help retail stores identify their customers. Karim stressed that the half life of AI knowledge is 2 years, so we needed to read and keep reading. Reading never ends. Sad story for my lazy self.

On Thursday morning, Dr Jacques Ludik of Cortex Logic took us on the concepts and theory of Deep Learning. We learnt the correlation and proper usage of the major buzz words: Artificial Intelligence, Deep Learning, Machine Learning, Reinforcement Learning, Natural Language Processing, CNN, RNN, and GAN. We learnt about the Deep Learning workflow. I discovered the application of Deep Learning in the healthcare space particularly in Drug Discovery and Disease Diagnosis(Breast cance detection with CNN).

Next we had Matt Graser of Bankable Frontiers Associate, USA teach us on Deep Learning for Financial Inclusion — Use of alternative data. Matt made a very important statement in this class saying, ‘Financial Inclusion is a means, not an end’ and that really hit home. It’s a tool for improving education, healthcare and the quality of life of low income people. This class dwelt a lot on CNN and using image detection to determine how much loans should be given to a trader.

Then Dr Emmanuel Doro’s class happened. The most exciting class of the entire bootcamp. Dr Emmanuel Doro is the Principal Data Scientist at Jet.com(owned by Walmart). Dr Doro took us on the basics of Convolutional Neural Networks, explaining the maths behind the Convolution, Pooling and flattening steps of CNN. Dr Doro made the statement that caught my attention. First is, ‘You should have a book to read every month’, explaining how he was able to solve problems at his company because of his reading culture. Second was, ‘If you’re not solving problems as a Data Scientist, you’re not a Data Scientist’. This last one pained my soul, he said, ‘If you don’t have a maths background, when everyone is taking 1 step, you have to take 3’. We learnt how to define a model using a Sequential class or a functional API, and lots more.

Next we had Dr Abiodun Modupe from the University of Wits, South Africa. Dr Modupe took us on the basics of Natural Language Processing. We created a neural network with 2 input layers, 2 hidden layers and 2 output layers. We then did several calculational computations on the forward propagation and backward propagation.

On Friday, we had Osayi Igharo of Ripple VC, USA. He taught us what investors look out for in AI Startups. He told us to ask 6 very important question. Why you? What’s your team like? Why now? How defensible is your product? How will you make money? Are you ready to be accelerated in the Silicon Valley?

Then we had the Recommender System class by our own Robert John, who’s the first Machine Learning Google Developer Expert in Sub Saharan Africa. Then we had a discussion session with the guys from Guardian Newspapers on the best ways to personalize news content and create a recommender system for their audience. This signals, that Nigeria is waking up to AI and it’s tremendous capacity and opportunities.

We also had a session with the team from Terragon Group, a presentation made by my friend, Lekan Wahab. Terragon gave us a challenge at the end of the presentation with an internship reward.

On Saturday, we had masterclasses with Sarah Hooker of Google AI and Dr Wale Akinfaderin. We had a session with the Co-founder of Bluechip Technologies on how to build a great technology company. Mr Yomi Akinyemi from KPMG spoke to us on what a consulting company looks for in a young analyst and its internship opportunity. Mr Kola Oni of Axa Mansard spoke on how AI can disrupt the insurance industry and we ended the day with a masterclass with Dr Omoju Miller of Github.

Sunday kicked off with Dr Jacques Ludik(who had the most classes at the bootcamp, teaching both DL and ML streams) taking, ‘Artificial Intelligence Applications in Real World: Global Trends and Best Practices’. Exciting exciting class after which we had breakfast, and then a panel session with Dr Abiodun Modupe and Dr Sulaimon Afolabi(both flew in from South Africa) on Academic Research in AI. Blessing Oladeji of Octave Analytics ended the bootcamp with a session on becoming an AI Entrepreneur. After which Dr Bayo Adekanmbi took us through the plans of Data Science Nigeria, going forward.

Key take away for me is the people i met from the Deep Learning stream and the awesomely productive discussion i had with folks like Mr Idris of Unilorin and Stephen Gabriel(a medical student from Uniben).

Altogether, it was a rich and exciting experience. Lots of fun and laughter. Lots of mathematics and mathematics. Lots of insights and thought-provoking revelation. I don’t know if i would forgive myself if i ever missed out on an opportunity like this.

Source: Deep Learning on Medium