Statistical change point detection of internet traffic
MPhys Astrophysics (2:1 St-Andrews)
I'M ORIGINALLY FROM
- Statistics - non-stationary processes(time series), dynamic graphical models, anomaly/change point detection
- Complex dynamical systems - agent based simulation, epidemiology of computer viruses
- Machine Learning - Semi/un-supervised (Clustering) and distributed reinforcement learning
- Networks - Mainly communication networks, internet Netflow and packet level transactions, also interested in social networking, community detection and recommender systems
- Physics(Theory/astro) - Crazy toy quantum gravity theories, conformal gravity, emergent quantum phenomena, measurement and geometry.
- Computer Science - Simulating botnets and learning to hack systems.
- Director at Social Enterprise ShareSci CIC developing a public science communication website.
- Worked for a bit at the UK Astronomy Technology Centre on the JWST/MIRI project, mainly instrumentation testing.
- Some research on epilepsy prediction at Oxford
- Helped testing and development of prototype optical interferometer at the Observatoire de La Cote d'Azur
WHY I APPLIED TO THE SECURITY SCIENCE DTC
My project was well funded, in London and something interesting..
WHAT I DO WHEN I'M NOT STUDYING
Work on developing ShareSci and my other ideas...but mainly eating, drinking and sleeping!
MY CAREER ASPIRATIONS
Apply my research and that of others within the field of behavioural change detection to problems across domains with specific interests in marketing, finance and security. This will probably take the form of some kind of research consultancy working on collaborative projects within the above industries.
WHAT KIND OF PARTNER ORGANISATION I'D LIKE TO WORK WITH AND WHAT I'D LIKE FROM THEM
My PhD is funded through the DstL, so hopefully have access to some awesome data sets and knowledgeable mentors. I would also be open to working with organizations from other industries where similar research methodologies can be applied...aka whoever has some interesting data to play with.
(In particular if you have high dimensional non-stationary continuous time series, and want to look at evolving correlation and group structure get in touch!!!)
LIST OF PUBLICATIONS
Large Scale Online Learning and Decision Making Workshop (LSOLDM 2013)
Young Statisticians Meeting(YSM 2013) - Talk and Poster: "Catching Cyber Attacks with a Lasso"
Dstl Big Data Workshop
Imperial College, Big Data Industry and Academia meeting
Recent advances on the changepoint problem
Dstl national PhD conference
Data Analysis for Cyber Security
alexander (d) gibberd (d) 12 (at) ucl (d) ac (d) uk
(d=., at =@)