Brown Bag Special: Deep Learning & Data Analytics

Earlier this month, Dr. Sadiq Sani, a Research Fellow at Robert Gordon University, Aberdeen dropped in on one of our Brown Bag sessions to give us a fascinating talk on Deep Learning & Big Data Analytics. We couldn’t let him go without a candid interview. Enjoy!sediqsani

M: Truly an honour to have you clear your schedule and come in to have a talk at FlexiSAF. Please tell us a bit about yourself. What’s your educational background?

Well, I have an undergraduate degree in Computer Science from ATBU. I graduated in 2006, same year as your CEO Faiz, who I have known from my undergraduate years. I then went on to do an MSc in Computing Information Engineering at Robert Gordon University in 2008. During my MSc project, I worked on a personalised video retrieval system which introduced me to the theory of information retrieval. This experience ignited the interest in me to do a PhD in a related area. I started my PhD at Robert Gordon University in 2010, working on semantic representations for text classification. During my PhD I worked on a number of machine learning algorithms and took interest in data science in general. I took a number of courses and attended summer school on data science. At the end of my PhD in 2014, I was employed by Robert Gordon University as  Research Fellow which is my current job.

M: Wow, quite impressive! Now the topic you discussed centered around Big data and analytics which seems to be all the rage today. Now, data analytics is often coupled with the term “predictive modelling”, a growing discipline of using data gathered in the past to predict what will happen in the future. Do you think this is the best move for Nigerian IT companies and why?

Well, data science is definitely the future, not just for IT companies, but for all industries in Nigeria, and across the world. The biggest direct beneficiaries of data science are companies that have plenty of (operations or customer) data, or companies operating in areas where a lot of relevant data is publicly available. Examples of such companies, in the Nigerian context are Banks, Telecommunications companies, Power distributions companies and  online retailers. Other beneficiaries are Government agencies and multinational corporations. All these industries can benefit from analysis of the data they have in their keep in order to derive insights and inform decision making.

M: I found the concept of deep learning fascinating but quite complex to grasp. Can you explain it a  bit better (in lay man’s terms)?

Not sure there is a way to actually explain deep learning in layman’s terms because deep learning is actually a complicated umbrella concept covering a number of different techniques, some of which are very complicated. But in simple terms, deep learning is an application of artificial neural network, which is a machine learning technique modelled after the human brain. The name ‘Deep learning’ comes from the network having multiple levels of abstraction which helps the algorithm to learn to make more complex predictions. Artificial neural network has had limited adoption in the machine learning community because of the the amount of computing power required to make it work. However, it has recently regained popularity because of the amount of computing power and training data we easily have at our disposal today.

M: Now I get it! A while ago, I read that Google’s TensorFlow is already offering a Deep Learning course, in collaboration with Udacity. The course is designed to help make deep learning even more accessible to engineers and data scientists at large. Do you think platforms such as these should be made more available in Nigeria and why?

Well, education is power and as Malcolm X said, ‘the future belongs to those who plan for it today’.  Thus, making tools that will help Nigerians learn and easily apply machine learning concepts is a very welcome idea. However, there is a need to go further and have machine learning being taught at our higher institutions as this will equip Nigerians with the ability to understand, in greater depth, the concepts behind these algorithms and enable them to tweak these algorithms to better meet the particular needs of the specific problem they are attempting to solve.

M: Finally, do you think Deep Learning is Nigeria’s future and why?

Machine learning in general has a future in Nigeria because this has the potential to produce machines that can accomplish tasks with higher accuracy, reliability and more at reduced cost, compared to using humans for the same task. However, the lack of infrastructure still remains a significant challenge that is likely to hamper progress in this regard. For example self-driving cars navigate roads safely by obeying street signs, traffic signals and road markings. These are virtually non-existent in most parts of Nigeria and thus, will definitely prove a major challenge for this type of technology.

M: Thanks again for taking the time to this!

It’s been an absolute pleasure, thank you!

 

Enjoyed this? Keep following the blog for more interesting Brown Bag sessions!

 

Comments

comments