Dr Xiaojun Zhai

Please explain your research in two sentences

I carry out research in the areas of embedded systems and system-on-chip design with a focus on optimising security, power, performance, and reliability for addressing the practical challenges in multidisciplinary research. Recently, one of my major research activities was to develop new adaptive deep learning hardware to support the deployment of artificial intelligence (AI) on resource-constrained embedded devices. This research would provide high performance and efficient AI solutions for new types of intelligent devices and anonymous systems in the home, workspace, and in extreme environments, of chips doing AI “everywhere”.

Why is your research important and what difference will it make?

Deep learning is the key technique in modern AI that has provided state-of-the-art accuracy on many applications. Today, although most of the computational loads of deep learning systems are still spent running neural networks in data centres, the ubiquity of smartphones, and the upcoming availability of self-contained wearable devices and autonomous robot systems are placing heavy demands on deep learning inference hardware with high energy and computing efficiencies along with rapid development of deep learning techniques. This is an exciting opportunity for the research of adaptive deep learning hardware at the edge, where the data is collected. Compared with traditional centralised methods, such local processing reduces communication cost, latency, and enhances privacy and reliability, and will have a huge impact on how AI will develop in the future.

Please provide a summary of your research achievements.

I have been involved as the key researcher/co-leader in a number of international and national research projects, including the National Centre for Nuclear Robotics (NCNR), and the Qatar National Priorities Research Program (NPRP). As a result of those research activities, some of my research work has been extended to develop novel security techniques for mobile and Internet of Things (IoT) devices for a successful start-up company (Metrarc Ltd), and the results from the CER2EBRAL project provided the first feasible solution to execute such large-scale software on resource-constrained embedded platforms, and was the first real-time local vasculature simulation enabled in the Qatar Robotic Surgery Centre. Recently, I have been awarded an EPSRC New Investigator Award (NIA), to help new academics in my field to develop their career. My research work to date has resulted in more than 85 scientific journal and conference publications, including IEEE/ACM Transactions and Journals, as well as several leading international conferences in my research areas.

What has been your biggest challenge and how did you overcome it?

My biggest challenge was to carry out high-quality research in suitable research environments to generate top publications, and secure funding from the mainstream funders. At the early stage of my academic career, my duty was mainly focused on education - I had very heavy teaching workloads and was only able to carry out my research in my spare time. However, fortunately, I collaborated with other world-leading researchers and received considerable support from them, so have managed to slowly move my research in the right direction. Particularly, when I joined CSEE at Essex, I received significant support and help from the research group and other senior members, which enabled my success and the biggest achievement in my academic career since I joined the school in September 2018.