The University of Southampton

Dr Tayyaba Azim 

Personal homepage

I am a computer scientist with expertise in Machine learning, Computer Vision and Natural Language Processing. I received my masters degree in Computer Science from the National University of Computer and Emerging Sciences, Islamabad, in 2009 and earned doctorate in Computer Science from the University of Southampton, Southampton, in 2014. Soon after completing my Phd, I worked as a Data Scientist with the Horizon Research Institute in University of Nottingham, and then as a Research Scientist at Cortexica Vision Systems, Imperial College of London, U.K. My first industrial assignment as a Research Associate in Horizon research institute was on the exploration of identifying customers’ behavioural trends via topic models, whereas my second industrial assignment as a Research Scientist in Cortexica Vision Systems included the deployment of deep models for improving image retrieval performance offered by the retailers (ASOS and Zalando). Both the jobs gave me an exposure to work on big multimedia data problems using deep learning models in the industry. This also paved my way to initiate research on natural language processing (NLP) tasks, as topic models tend to mine abstract themes in a collection of documents. I further pursued my research in this direction on joining academia as an Assistant Professor of Computer Science.


Research interests

My doctoral research highlighted the possibility of deploying deep learning models for improving the classification performance of state of the art kernel methods like support vector machines. The research showed how such a hybrid approach can combine the best of both the paradigms for computer vision problems. The research experience gained from my doctoral program cultivated and nurtured lifelong skills of working on daunting ideas that can create a difference. From the last six years, I have extended this research by focussing on developing Fisher kernel methods that can bridge the gap between the two popular frameworks: Deep learning and Kernel methods. This research has helped me in winning several national and international research grants from the industry and academia. I am recipient of a Startup Research Grant, a National Grassroots ICT Research Initiative Fund, a National ICT Research and Development Grant, and an have received the Best Paper Award at ICPRAM, in 2017.

 I have also worked as a technical reviewer of the following journals, conferences and organisations: IEEE Transactions on Neural Networks, IEEE Access, IET Electronics letters, Neural Processing letters, Journal of Information Sciences. Journal of Pattern Recognition, IGNITE National ICT R&D, Pakistan, IEEE International Conference on Emerging Technologies  (ICET) and IEEE International Conference on Industrial and Information Systems (ICIIS).

My current research interests include deep learning methods for graphs in NLP. Im also interested in self-supervised learning methods for deep models to develop intelligent chat bots. Such learning techniques are useful for online lifelong and continual learning, where most of the encountered real world data is unlabelled. Zero shot and one shot learning techniques are also of significant interest to me in this regard.


I have experience of teaching in the Higher Education sector both in the UK and abroad in Pakistan. This experience has enriched my knowledge of cultural and societal differences, crucial elements to promote diversity in a learning environment. Under my recent occupation as an Assistant Professor at the Institute of management Sciences, I have served various roles as a course instructor and manager industry academia linkages. As an inclusive practitioner the new courses I introduced in campus are: Data Science, Multimedia Databases and Machine learning. Besides delivering lectures, I was actively involved in curriculum development and have applied various learning strategies to engage all students in an effective teaching and learning environment and not focus on a specific group or diversity dimension.


Azim, Tayyaba, Loitongbam, Gyanendro Singh and Middleton, Stuart (2022) Detecting moments of shange and suicidal risks in longitudinal user texts using multi-task learning. Workshop on Computational Linguistics and Clinical Psychology: North American Chapter of the Association for Computational Linguistics 2022 (NAACL-2022), , Seattle, United States. 15 Jul 2022. (In Press)

Share this profile FacebookTwitterWeibo