Andrew Elliott

Network Scientist

Postdoctoral Researcher

The Alan Turing Institute

Email: aelliott AT turing.ac.uk

Personal Email: ande.elliott AT gmail.com

Google Scholar Turing Page

CABDyN Page ORCID Page

I am a postdoctoral researcher at The Alan Turing Institute, focusing on Network Science and supervised by Prof. Gesine Reinert and Dr Mihai Cucuringu. I received my D.Phil. from the University of Oxford with a thesis looking at subsampling in biological networks, under the supervision of Prof. Felix Reed Tsochas and Prof. Gesine Reinert.

More specifically, I focus on developing scalable network-based algorithms to solve various real-world problems, such as anomaly detection and core-periphery detection. This includes improving and extending existing methods, developing new methods to solve a wider class of problems, and applying well-understood methodologies to gain insights in new application areas. Recently, I have also collaborated on a number of machine learning projects.

I am interested in applying these algorithms to a wide variety of domains, including biology, social sciences and physical sciences. Most recently, I have worked on several projects looking at applications in financial systems, cybersecurity and urban networks.

Research

Postdoctoral Project

The Alan Turing Institute, August 2017 - Present

Supervised by Gesine Reinert and Mihai Cucuringu.

Initially a 2-year, industrially-sponsored position funded by the Accenture Turing Alliance, extended by one additional year to collaborate with further industrial partners.

This involved working in close collaboration with industrial partners, including weekly progress reports and agreeing academic and business direction of the project, creating attractive demos and presenting them to potential clients.

Main project focused on anomaly detection in networks, using network comparison and spectral methods. This involved adapting known network comparison and spectral methods and constructing new methods to score anomalies in networks.

Further work included working on a project extending core-periphery structure to directed networks. This involved creating methods to find the structure, testing said methods and validating the approach on real-world data. We are currently in the process of applying our new core-periphery methods to our anomaly detection pipeline.

Selected Ongoing Projects/In Preparation

Developing a network comparison library

Work with academics and the Turing research software engineering team to develop a network comparison library in R.

https://github.com/alan-turing-institute/network-comparison

Urban Street Networks

Work with Stephen Law and Luis Ospina on analysing street networks. Results presented at NetSciX.

Motif Based Spectral Clustering

Joint project with William Underwood and Mihai Cucuringu on motif clustering, including fast routines to compute motif adjacency matrices.

https://arxiv.org/abs/2004.01293

Computer Vision: Explainability and adversarial examples

Work with Chris Russell and Stephen Law on explainability and adversarial examples using perceptual loss.

https://arxiv.org/abs/1912.09405

Previous: D.Phil. (PhD) University of Oxford

SABS-IDC Doctoral Training Centre, CABDyN Complexity Centre, Said Business School

Thesis title: Path-based sampling and community detection methods for biological and other applications

Thesis available here

Supervised by Felix Reed-Tsochas, Gesine Reinert (Department of Statistics, CABDyN Complexity Centre), Elizabeth Leicht (formerly CABDyN Complexity Centre, currently Data Scientist at Facebook), Alan Whitmore (e-Therapeutics plc).

Examiners: Dr. Mariano Beguerisse (Internal), Prof. Mark Newman (External).

Project in collaboration with a e-Therapeutics plc

Project Focused on:

  • Subsampling Problems in Networks
  • Community Detection in Sampled Networks
  • Network Null Models

Courses attended throughout DPhil:

  • Systems Biology Doctoral Training Centre (Biology Overview)
  • Probability and Statistics for Network Analysis
  • Networks
  • Statistical Methods

Papers

  • A. Elliott, E. Leicht, A. Whitmore, G. Reinert and F. Reed-Tsochas. A Nonparametric Significance Test for Sampled Networks. Bioinformatics https://doi.org/10.1093/bioinformatics/btx419
  • A. Elliott, A. Chiu, M. Bazzi, G. Reinert and M. Cucuringu Core--Periphery Structure in Directed Networks Under Review https://arxiv.org/abs/1912.00984
  • A. Elliott, S. Law and C. Russell Adversarial Perturbations on the Perceptual Ball Under Review https://arxiv.org/abs/1912.09405
  • Underwood W. G., Elliott A., Cucuringu M. Motif-Based Spectral Clustering of Weighted Directed Networks Under Review https://arxiv.org/abs/2004.01293
  • Mas, I. and Elliott, A., Where's the Cash? The Geography of Cash Points in Tanzania (Nov 11, 2014). Financial Sector Deepening Trust (FSDT) Focus Note, No. 2, 2013. Available at SSRN: http://ssrn.com/abstract=1875883
  • A. Elliott M. Cucuringu, P. Reidy, M. Luaces and G. Reinert Anomaly Detection in Networks with Application to Financial Transaction Networks https://arxiv.org/abs/1901.00402

In Preparation

  • A. Elliott, M. O'Reilly, L. Ospina Forero, G. Reinert, A. Wegner Netdist: Network comparison in R. For further details see here
  • S. Law, A. Elliott and L. Ospina-Forero Characterising road networks through subgraph graphlet analysis
  • A. Elliott, E. Leicht, G. Reinert and F. Reed-Tsochas. Community Detection in Path Sampled Networks


Talks

  • Conference talk at NetSciX 2020, Core-Periphery in Directed Networks
  • Conference talk at NetSciX 2020, Characterising road networks through subgraph graphlet analysis (Constructed jointly and given by Stephen Law)
  • Conference talk at Complex Networks 2019, Core-Periphery in Directed Networks
  • Invited talk - Workshop: Network Science in Financial Services The Alan Turing Institute (Dec. 2019), Anomaly detection and core-periphery structure in networks.
  • Invited presentation at FCA (Financial Conduct Authority) (February 2019), Anomaly Detection in Networks using Spectral Methods and Network Comparison Approaches
  • Conference talk at COSTNET 2018, Anomaly Detection in Networks
  • Presented at the 2018 Flying Money Conference in Amsterdam - Also included in the proceedings available here: https://issuu.com/instituteofnetworkcultures/docs/flying_money_conference_reader_2018
  • Conference talk at COSTNET 2017, Anomaly Detection in Networks
  • Presentation to the Networks Journal Club (Feb 2016) Community Detection in Path Sampled Networks
  • Invited speaker at the Networks Pharmacology Workshop at University of Oxford (15 September 2015), A Nonparametric Significance Test for Sampled Networks
  • Invited seminar at the Centre for Networks and Collective Behaviour at Bath University (6 February 2014), Selecting Subnetworks for Further Study using a Novel Null Model
  • Talk at Doctoral Training Centre 4th Year Symposium (January 2014), A Systematic Study of Parkinson's Disease Network Construction through Sub-sampling and Seed Lists
  • Contributed talk at the Mathematics Of Networks meeting (16 September 2013), A Novel Null Model to Test the Significance of Features of Sampled Networks


Posters

  • COSTNET 2019 - Detection of Core-Periphery Structure in Directed Networks - Won Poster Prize
  • Complex Networks 2018 - "Anomaly detection in networks using spectral methods and network comparison approaches"
  • NetSci 2013, A Novel Null Model to Test the Significance of Features of Sampled Networks
  • Cambridge Networks Day 2013, A Novel Null Model to Test the Significance of Features of Sampled Networks
  • EMBL Student Symposium 2012, A Systematic Study of Parkinson's Disease Network Construction through Sub-sampling and Seed Lists
  • Systems Biology Doctoral Training Centre Conference 2012, Protein Interaction Networks and Snowball Sampling - A Good Mix?


Industrial Presentations:

Worked with a team to construct an interactive demo to present to selected industrial partners and industrial conferences.

International Compliance Association - The True Cost of Financial Crime

Flying Money Conference 2018 Amsterdam

Software:

Network Comparison Code:

Network comparison package code, developed by the The Alan Turing Institute's research software engineering team (notably Martin O'Reilly) with collaboration and contributions from various researchers:

https://github.com/alan-turing-institute/network-comparison

Paper: Anomaly Detection in Networks with Application to Financial Transaction Networks

Software can be found here, paper can be found here.

Paper: Core--Periphery Structure in Directed Networks (paper is available here)

Software can be found here. paper can be found here.

Paper: Motif-Based Spectral Clustering of Weighted Directed Networks

Software can be found here, paper can be found here.

Paper: A Nonparametric Significance Test for Sampled Networks

Software can be found here, paper can be found here

If you have any queries about the software please email on the address above.

Additional Functions for the DTC Networks Course

The NetworkX library is missing a small number of functions needed for the DTC Networks course. An additional file with the needed functions can be found here

Service

Workshops

Co-organised two workshops:

Reviewing papers

Reviewed/Co-reviewed for:

  • International Conference on Complex Networks and their Applications
  • Association for the Advancement of Artificial Intelligence
  • IEEE Computer Society
  • IEEE Intelligent Systems
  • Applied Network science


Research Software Working Group - The Alan Turing Institute

  • Representing the academic viewpoint for development of computing services at The Alan Turing Institute.