Sevvandi Kandanaarachchi

Research Fellow

Department of Econometrics & Business Statistics

Monash University, VIC, Australia

email : sevvandi.kandanaarachchi@monash.edu

I am an inter-disciplinary researcher working on anomaly detection, meta-learning, time series analysis, statistical and machine learning related research projects. I also like to conduct research on real-world problems. Before my interest sparked on statistical learning, I conducted research in geometric analysis, which is a branch of pure mathematics. In addition to research, I have also taught many courses including linear algebra, calculus, numerical methods, discrete mathematics and introduction to dynamical systems.

I obtained a BSc.Eng with first class honours from the University of Moratuwa, Sri Lanka in 2002. In 2006, I completed a MSc. Preliminary in Mathematics from Monash University and was awarded IPRS & MGS scholarships for my PhD. I obtained my PhD in 2011 in Mathematics also from Monash University. In 2015, I completed a Graduate Certificate in Data Mining and Applications from Stanford University.

Research Projects

1. Anomaly detection in streaming data (2017 – ongoing)

Anomalies tell a different story from the norm. For example, Mercury’s anomalous motion was not explained by Newton’s laws but was explained by Einstein’s theory of general relativity. Some examples of anomalies nearer to earth include fraudulent credit card transactions amongst billions of legitimate transactions; fetal anomalies that may indicate severe disabilities of unborn infants; chromosomal anomalies in tumours heralding cancers such as leukaemia; unusual trends or patterns in social media that may herald social unrest or terror activities.

The challenges increase in number and severity when we consider data streams. This project focuses on anomaly detection in streaming data (link). This is ongoing work with Prof. Rob Hyndman and Prof. Kate Smith-Miles.


2. Vegetation detection and bushfire safety project (2017 – ongoing)

This project aims at reducing/preventing bushfires caused by branches falling on powerlines. The project started off as a response to the “Vegetation Detection Challenge”which was an initiative of the State Government of Victoria, Australia. The purpose of this project is to reduce the risk of fire caused by vegetation on powerlines.

This is joint work with Dr. Mario A. Muñoz Acosta at University of Melbourne and Ms. Dilini Talagala at Monash University. We were awarded the second prize in the competition. Link


3. Space-Debris project (2015)

This project considered the regular bombardment of space-debris upon space vehicles and the damage caused by the impact of the collision. The objective of the project was to predict the severity of damage caused by the space debris.

This project was joint work with Prof. Kate Smith-Miles (then at Monash University) and Dr Shannon Ryan of DST Group.


4. Mean curvature flow (2007-2011, 2018 - ongoing)

Visualise an egg shrink in such a way that the more curved regions shrink faster, and the less curved regions shrink slower. Loosely speaking, this is mean curvature flow. More precisely each point moves inward at the speed of the curvature of that point. If the egg shrinks while preserving its enclosed volume, it is called volume preserving mean curvature flow. We are interested in the partial differential equations that result from these flows. This research was conducted during my PhD candidature at Monash University under the direction of my PhD supervisor Dr. Maria Athanassenas. I have restarted working on it with Dr. John Head.