Profile
After obtaining the Ph.D. in Electronic Engineering from University of Surrey in 2011, Varuna worked as a work package leader of a Large Scale Integrated Project "ROMEO" funded by the European Commission. During the time he was with University of Surrey, both as a Ph.D. student and as a research fellow, he developed a highly significant publication record in the area of image and video processing, which includes 12 highly cited IEEE Transactions, and an IEEE consumer electronic society award.
To pursue a career in the industry, in November 2013, he joined Apical Ltd, where he worked as a new algorithms developer for imaging and display systems to be run on efficient hardware platforms. The algorithms he developed were realized in nearly 500 million devices around the globe, and a patented technology for CMOS cameras, for which he is the lead inventor which appeared in a significant proportion of CCTV cameras in the world.
Since May 2016, he assumed duties as a lecturer in digital engineering at Loughborough University, where he was promoted to a Senior Lecturer in Jan 2020. He is the module leader for Principles of Data Science, Bayesian Methods in Deep Learning, Statistical methods in Finance, and Gaming Technologies and Simulation. He is the recipient of two EPSRC grants in the area of next generation machine intelligence, and complementary industrial funding. He currently has active industrial collaborations with several organizations. His team works on Engineering application of Artificial Intelligence, specifically Multi-Agent Reinforcement learning, multimodal computer vision, and simulation models, with a Vision to scale state-of-the-art AI algorithms to real world Engineering systems.
Varuna is the Programme Director of Artificial Intelligence and Data Analytics MSc programme at Loughborough University London.
Academic background
B.Sc. (Hons) Engineering, University of Moratuwa, Sri Lanka, 2007, First Class GPA 3.96
Ph.D. Electronic Engineering, Centre for Vision Speech and Signal Processing, University of Surrey, 2011
- Overseas Research Scholarships Award, HEFCE UK, 2008
- IEEE Chester Sall Award, Consumer Electronics Society, 2010
- Vice Chancellor's Award for Best Post Graduate Research Student, 2011
PG Cert in Academic Practice, Loughborough University, 2018
Lecturer, Loughborough University, 2016 - 2020
- Programme Director of Digital Marketing MSc, 2018 - 2020
- Programme Director of Artificial Intelligence and Data Analytics MSc, 2021
Research
Current research and collaborations
Sensor data perception in driverless vehicles: Dr De Silva, along with his PhD students, work on various aspects of sensor data processing in driverless vehicles. He is particularly interested in cognitive approaches to learning from data, to deal with unpredictable situations. His work includes sensor data fusion (LiDAR/Imaging/Radar), free space detection, cognitive approaches to object detection and control.
He is also working on player performance prediction in a collaboration with Chelsea Football Club.
Areas of research expertise
Varuna works on the following fundamental areas of research in the field of Machine Learning. While his work is focused on the fundamental AI Developments, Varuna's team work mostly on Sports analytics and Driverless Vehicle Control applications to apply this research.
- Multi-Agent Reinforcement Learning: Building intelligent agents using reinforcement learning in multi-agent settings. Specific interests are in scalable reinforcement learning and methods to in build human domain knowledge and experience in to the reinforcement learning framework. The applications include, team sports analytics, data driven ghosting, mobile edge intelligence, vehicular traffic control.
- Deep generative modelling: Learning multimodal representations using deep neural networks such as Variational Auto-encoders and Normalizing flows. Applications include: text to image synthesis, neuro-symbolic approaches for sports analysis, and scene recognition, Portfolio optimization, customer profiling in finance
- Neuro-symbolic approaches for computer vision: using logical representations in conjunction with neural networks for explainable and transferable machine learning model development. Applications include: body pose estimation, human activity recognition, and scene recognition in driverless vehicle control
Current PhD / research supervisions
Current PhD supervisions as a Principal Supervisor:
- Joosep Hook: Multi-agent Reinforcement Learning, Start: July 2018 - August 2021 (Intended Thesis Submission) -
- Haylat Tibebu: Neuro-Symbolic Computer Vision for Driverless Vehicle Control, Oct 2018 -
- Jamie (Jiaming) Na: Computer vision for Sports Analytics, : April 2020 -
- Rafael Moreira-Pina: Scalable Multi-Agent Reinforcement Learning, Start Oct 2020 -
- Aaron (Chaoyi) Gu: Multimodal Imitation learning for Football player modelling, Start Oct 2020 -
- Ben (Yisheng) Pei: Deep generative modelling of event and tracking data in football for probabilistic tactical analysis, Start Jan 2021 -
- Elim (Yilin) Feng: Symbolic approaches for transferable deep learning, Start March 2021
Current PhD supervisions as a Co-Supervisor:
- Haoyang Qin: Hybrid modelling of flood dynamics, Start Jan 2020, Principal Supervisor: Prof Qiuhua Liang
- Eyad Alsaghir: Transferable models for AI in Engineering Applications, Start July 2019, Principal Supervisor: Prof Xiyu Shi
Completed PhD / research supervisions
- Dr Jamie Roche: Principal Supervisor - Multimodal Perception in Driverless Vehicle Control, Start Oct 2016 - End Sep 2019, Currently Lecturer, Institute of Technology Sligo, Ireland
Interests and activities
- Fellow of the Higher Education Academy
- Member of IET
- Member of BCS
- Associate Editor of IEEE Access
- Associate Editor of MDPI Vehicles
- Pillar Lead - Capability Development - AI Centre for Excellence at SLASSCOM Sri Lanka
- Governor of St. Paul's Secondary School in Leicester