Research Interests and Current Research

My research direction is to develop novel anomaly detection approaches in the internet of things. Substantial work has been done in this area; however, they are based on idealised settings. My work is to develop new statistical theories and methods to identify anomalies in IoT settings accurately and immediately.

Besides, I have a strong interest in applying statistical methods to solve real-world problems.

Conference/Workshop:

  • International Conference on Computational Statistics (COMPSTAT2024) (August 2024)
    • Oral: A fast Bayesian online changepoint detection algorithm
  • European Meeting of Statisticians (July 2023)
    • Oral: Detecting changes in a distributed system in real-time with unknown parameters: from Gaussian to mixed-type data
  • Paris-Saclay Change Detection Workshop (January 2023)
    • Oral: Online distributed changepoint detection methods for monitoring the Internet of Things
  • Research Student Conference in Probability and Statistics (September 2022)
    • Oral: An online distributed changepoint detection method for monitoring the Internet of Things

Event:

  • Institute of Mathematics and its applications webinar (June 2024)
  • TakeAim 2024 awards ceremony – runner-up (March 2024)
    • Presentation: Smarter, Safer and More Sustainable Internet of Things Monitoring
  • Florence Nightingale Day (Jan 2024)
    • The Florence Nightingale Day will showcase successful women in mathematics at various stages of their careers, display information about the broad range of possibilities offered by a degree in mathematics or statistics, (see more information: https://www.lancaster.ac.uk/maths/engagement/working-with-schools/florence-nightingale-day/).
    • Presentation: My journey: exploring the fascinating world of statistics”.
      • Have you ever thought about what makes a movie really popular at the box office? Does having more money make people happier? And how exactly do viruses spread across different places and over time? These are all projects I worked on during my undergraduate and postgraduate studies. They sparked my curiosity to explore statistics further. Now, as a PhD student, I work on detecting unusual behaviour in data streams: the challenge is to do this as quickly as possible. In this talk, we will explore these projects and I hope to show you how enjoyable statistics can be!
  • Lancaster University summer school (August 2023)
  • Steps to University for Mathematics Students (May 2023) – aimed at Year 12 girls
    • Presentation: Exploring the Fascinating World of Statistics (https://amsp.org.uk/our-sums-steps-to-university-for-mathematical-students-enrichment-day-is-back-for-2023/)

Teaching Assistant

2023-2024

  • MATH336: Machine Learning
  • MESME (Mathematics Education for Social Mobility and Excellence) mentor: supports students from all backgrounds to achieve mathematical excellence so that they feel more confident as mathematicians and go on to have a greater range of future personal and employment choices. (see information: https://mesme.org/)

2022-2023

  • MATH104: Statistics
  • MSCI212: Statistical Methods for Business
  • MATH331: Bayesian Inference

2021-2022

  • MSCI212: Statistical Methods for Business
  • MSCI562: Intelligent Data Analysis and Visualisation
  • MSCI580: Analytics in Practice

Dissertation

  • Investigating the effects of temporal biased sampling and population temporal trend on a phylogeography method based on the structured coalescent

Statistical phylogeography is a statistical method which is based on the Full Bayesian method and Continuous Time Markov Chain, which could be used to find out the origin location and time, and the transmission route of the virus. Temporal biased sampling is introduced by inappropriate sampling method, and population temporal trend is a common phenomenon in evolution. By simulating four different scenarios, we evaluated the effect of MASCOT (which is a phylogeography method based on the Forward-Backward algorithm) under such situations. Thus, we draw a conclusion that MASCOT is not suitable to analyse genetic data having temporal biased sampling.