Could you tell us about your background? 

Born in Mauritius, a tropical Island, I did my undergraduate studies in Physics at the University of Mauritius. I was then awarded the prestigious Square Kilometre Array (SKA) Masters bursary and embarked upon an MSc in Astrophysics and Space Science at the University of Cape Town and North-West University. In April 2018, I started my PhD studies in the Jodrell Bank Centre for Astrophysics (JBCA) at the University of Manchester, under the supervision of Prof Benjamin Stappers and Dr Rob Lyon.

 

Please tell us about your PhD research.

We cannot escape the fact that astronomy is already in the era of Big Data, with various surveys generating from petabytes up to exabytes of data per day. Sifting through this gigantic amount of data will be almost impossible for astronomers. In the same spectrum, the crux of my PhD lies at the intersection of data science and astronomy; my research is mostly in the development of a sophisticated and automated pipeline for transients detection and classification. In particular, my work involves the combination of unsupervised methods for feature evaluation and supervised methods to classify variable stars (time-series data). In addition to this, one of the most prominent issues in science and particularly in astronomy is imbalanced data. My research also focuses on developing various algorithms to augment datasets which will prevent overfitting when implementing machine learning (ML) algorithms, thus alleviating the imbalance problems. During my PhD I have developed a hierarchical classification pipeline that performs classification and augments data level-wise in the taxonomy at the same time. Currently I am currently working on an automated pipeline to perform the classification of transients and bogus candidates using MeerLICHT data. Additionally I am building an algorithm to detect and localise Fast Radio Bursts (FRBs) using convolutional neural networks.

​

More details about my research can be found at zafiirah.github.io

 

​

​

​

​

​

​

​

​

​

​

​

​

 

 

 

 

 

 

 

 

 

 

 

What started your interest in astronomy and data science? 

In 2015, I attended the Mauritius Machine Learning JEDI workshop, organised by Prof Bruce Bassett and Dr Nadeem Oozeer.  JEDI style workshops are more open to solving research problems, rather than the format of long formal talks, and they also encourage more interaction between participants and senior academics. This helped me to have a broader view on the intersection of astronomy and machine learning. The project that I was working on was source classification in radio astronomy using machine learning. It was during this time that I realised just how powerful data science techniques can be in astronomy. The workshop was 10 days long and so we did not have the time to reach our end goal; this interesting project later formed the research for my Masters thesis in 2016. I had the opportunity to apply a Generative Adversarial Network to simulate radio images and perform classification of real sources.

 

What would your dream job be, and where?

I strongly believe that AI can have a significant impact on real world challenges, and has the potential to bring insightful solutions to various societal problems. After my PhD, I have a strong zeal to work on predictive and analytical modelling of epidemic outbreaks, predetermining risk areas using geospatial data and notifying people in impacted areas, and also to provide insights on flood prevention. I would like to work in one of the Big Tech companies e.g. Google AI, DeepMind, Facebook AI, Microsoft AI or IBM AI Research amongst others.

​

What accomplishment are you most proud of so far?

Having the opportunity to pursue a PhD in the United Kingdom is one of my greatest accomplishments so far. I had never imagined myself going so far in Mauritius given the limited opportunities and infrastructures. In Mauritius, it is very rare to find women pursuing further studies abroad especially. I am thankful to my family and to DARA Big Data for providing a programme that encourages women and under-represented countries to dream big.

​

What advantages do you think there are for students with machine learning skills, particularly in Africa?

Machine learning is the new fuel which makes the most efficient use of data, enabling us to draw useful inferences. I believe students with ML skills are greatly valued given the large rich data sets that are gathered with technology. ML skills are so broad that they can be applied in various domains, from computer and life sciences to the financial and business sectors. Therefore there is a huge opportunity to work on real-world problems and develop solutions that will have a deep impact on all walks of life.

  

What have you enjoyed most about the UK while you've been studying here?

I have had the opportunity to attend many prestigious conferences, e.g. ICLR, ICML, OxML, WiML and Google Sandbox among others, which would not have been possible previously.