Network Comparison

Lead Research Organisation: University of Oxford
Department Name: Statistics

Abstract

Networks are ubiquitous, a prime example being social network sites such as Facebook, and yet there is a lot which is not understood about them. Even for the seemingly simple question of how similar two networks are, to date there is no generally accepted method available to answer this question.

The proposed project will address exactly this issue, how to compare networks. Network comparison is of considerable importance for example when trying to compare protein-protein interaction networks of organisms under different form of stress, or when trying to compare protein-protein interaction networks of people carrying a certain disease to that of people without the disease.

The statistics will be applicable to any types of networks, and they could be used to track changes in networks over time.
While constructing such a statistic for network comparison may be relatively straightforward, for statistical tests its probabilistic properties have to be understood. Achieving such understanding requires considerable expertise in probability, as there are a number of questions which have to be tackled in order to establish a useful asymptotic result.

Similarly, the statistical properties of the new statistics have to be understood. How well do they separate models which are generated from just slightly different models? How can suitable test procedures be implemented?

Preliminary studies have shown that our method can be used to reconstruct phylogenetic trees based on protein-protein interaction networks. If confirmed, then this is the first result which shows that the topology of protein-protein interaction networks alone contains information about evolution. This result is of considerable interest in biology, and its biological implications have to be studied in detail.

Experience has seen that networks are able to capture the imagination, and are a suitable topic for outreach activities. Hence we intend to develop some lectures aimed at a general audience, to be offered to schools, via departmental contacts in the first instance, as well as at public science forums.

There will be a blog about our research on the Oxford Sparks portal, an on-line public science website.

Planned Impact

The most striking impact of the research could be on our view of evolution; protein-protein interaction networks contain information about evolution in their topology.

The research will be of immediate interests to companies like e-Therapeutics who specialise in network pharmacology. There is an existing collaboration between Gesine Reinert and Alan Whitmore from e-Therapeutics on Parkinson's disease. The method would allow comparing different networks which are conjectured to represent the disease, and would help identify drug targets. Charlotte Deane collaborates with Matt Page and Colin Stubberfiled at UCB - the biopharma company on how networks change over time and between different disease and healthy states.

The potential applications are wide-ranging. One could compare the communication networks of different NHS trusts to assess whether there is a correlation between efficiency and network topology. Gesine Reinert has discussed related questions with Dr Kazem Rahimi at the George Centre for Healthcare Innovation.

Network comparison could shed light on trade behaviour through the comparison of trade networks across geographical regions and commodities. These results could in turn inform international development policies.

The research would advance the scientific areas of applied probability, statistics, biology, social science, computer science and the emerging field of network science, through a new method, new approaches, and new results in specific applications.

The project would train two postdoctoral researchers as well as project students. The results would be disseminated to the wider public via lectures and a blog, and they may motivate young people to take up studies in mathematics and statistics.
Finally, our results may help establish the emerging discipline of Network Science as a discipline with its own research questions and tools, which go beyond the research questions raised in particular applications.

Publications

10 25 50
 
Description We have developed two tools to compare networks, with the aim to find local patterns in networks; they are called Netdis and NetEMD. We have also extended network tools to include the case where nodes in networks have certain characteristics, such as damage caused when the node is hit by a disaster.

We have also developed a subsampling algorithm for Netdis which allows comparing networks even when the networks are not fully known. This algorithm can be used to compare large networks such as websites linking to other websites; the exact number of websites does not have to be known in order to carry out the comparison. The algorithm is of interest to other fields as it is based on bootstrap on networks. We not only give the algorithm but we also prove that it converges to the correct outcome as long as the graph is sufficiently homogeneous, while allowing for some heterogeneity.

Our second method, NetEMD (the paper is currently under revision), compares networks directly through the distribution of features of the networks, such as subgraph counts. The use of features also appears in classification of networks through machine learning kernels. NetEMD performs as well as the best-performing machine learning kernel methods we tried, without requiring any additional information beyond the network topology.

To understand what to expect when comparing networks, we carried out some thorough mathematical analysis of underlying models. In insurance networks which relate catastrophic risks we determined two key indices which describe the behaviour of the system depending on the underlying networks. These indices can be used to monitor markets and also to regulate markets.

We are still in the process to find out why our methods work so well on biological data - do they capture some evolutionary principle? It turns out that our network comparison methods give slightly different results on newer networks. We explored this difference by separating the protein-protein interaction data according to the experimental method used to obtain the data. For one method (TAP-MS) the network comparison methods do not work very well, whereas they perform well on the other main method (Yeast-two-hybrid). There are good underlying reasons for this difference, the two methods produce different types of errors and are biased towards different types of proteins. Based on these insights we have started building an evolutionary model for protein-protein interaction networks.
Exploitation Route Our method was used to compare ecological networks, by Drew Purves and his group when he was at Microsoft Research, Cambridge. Sarika Jalan and her group use it to analyse networks arising from fMRI data. We understand that GCHQ has been using our methods. Our methods form the basis of a 2-year project with Accenture on threat detection in networks.

The network comparison methods are almost fast enough to give real time network comparisons; we are trying to make them faster so that they can be used for detecting anomalies in networks, with possible application in cyber security. The methods will be developed to yield fine-grained information about locations of changes; which node combinations are the ones which trigger the network comparison statistic to be larger than expected? The theoretical underpinning will rely on a framework for comparing multivariate discrete distributions.
Sectors Communities and Social Services/Policy,Environment,Financial Services, and Management Consultancy,Pharmaceuticals and Medical Biotechnology,Security and Diplomacy

URL http://www.blopig.com/blog/author/anatol-wegner/
 
Description Our network comparison methods use counts of small subgraphs in a network as features which can be used to classify networks and to test whether networks come from a particular distribution. These count statistics are local summaries and hence can be updated efficiently. Moreover a detailed analysis of the network comparison statistic reveals the main local differences between networks through their contribution in the network statistic. This work lies at the foundation of a 2-year project with Accenture on threat detection in networks, with particular emphasis on anti-money laundering and modern slavery. This local refinement is also of interest both for our collaborators in network pharmacology and in cyber security, as follows. Suitable network comparison methods are key to identify subnetworks of protein-protein interaction networks which are relevant to a certain disease. In order to achieve more impact a fine scale analysis is now required so that particular protein combinations can be identified which distinguish healthy state from disease state. Through the funded project our collaboration with e-Therapeutics has intensified and now we are carrying out a detailed network analysis using scores on interactions, to detect small parts in a network which may change in the disease state. We anticipate a sandpit meeting with e-Therapeutics and interested researchers to take place before April 2018, funded through the COSTNET grant. Gesine Reinert has also been collaborating with Prof Niall Adams (Imperial) to test our network comparison methods, combined with his change point methods, on GCHQ data sets. We expect to submit a grant proposal in the near future to explore this method further. Our theoretical results have attracted interest from insurance companies and regulators; indeed we were awarded a Science of Risk Prize by Lloyd's of London. The Science of Risk Prize has increased the visibility of our work, and we have had conversations with representatives of the Bank of England about regulatory interventions to control systemic risk in insurance networks. Network comparison also plays a role in a follow-up project, funded by Novo Nordisk, on identifying modules in liver protein interaction network data under different conditions. The project provided material for 3 M.Sc. projects, 3 4th-year projects, one summer student project, and one visiting student project. So far 5 publications arose from these projects, with one more paper submitted. Perhaps the most important impact is the impact through our outreach activities. Networks are a good subject to explain to A-level pupils, and we have enjoyed talking to them. Our talk on networks is now part of the departmental repertoire for open days. It is also useful to have the blogpost on our research; we point to it regularly.
First Year Of Impact 2014
Sector Communities and Social Services/Policy,Financial Services, and Management Consultancy,Pharmaceuticals and Medical Biotechnology,Other
Impact Types Societal,Economic

 
Description Alan Turing scoping meetings
Geographic Reach National 
Policy Influence Type Participation in a advisory committee
Impact Gesine Reinert was invited to two scoping meetings for the Alan Turing Institute. At these meetings she talked about network analysis and its applicability for high-dimensional data sets. There were about 30 scoping meetings, and only 5 themes were eventually selected as focus points for the Alan Turing Institute. Network analysis is one of them. Subsequently Gesine Reinert was elected as an Alan Turing Faculty Fellow and is regularly called upon to discuss network issues, both as training of early career researchers and in a consulting role for companies such as Accenture.
URL https://www.turing.ac.uk/
 
Description Network blog
Geographic Reach National 
Policy Influence Type Influenced training of practitioners or researchers
URL http://www.blopig.com/blog/author/anatol-wegner/
 
Description ARC Discovery Funding
Amount $775,728 (AUD)
Funding ID DP150101459 
Organisation Australian Research Council 
Sector Public
Country Australia
Start 01/2015 
End 12/2017
 
Description Alan Turing small grant scheme
Amount £18,000 (GBP)
Organisation Alan Turing Institute 
Sector Academic/University
Country Unknown
Start 12/2016 
End 03/2017
 
Description COST
Amount € 126,000 (EUR)
Organisation European Union 
Sector Public
Country European Union (EU)
Start 06/2016 
End 05/2019
 
Description NSF grant - Co-PI
Amount $202,929 (USD)
Organisation National Science Foundation (NSF) 
Sector Public
Country United States
Start 09/2015 
End 08/2018
 
Description Novo Nordisk Pump Priming
Amount £20,750 (GBP)
Funding ID BRD00050-AM00 
Organisation Novo Holdings A/S 
Start 10/2018 
End 03/2019
 
Description Programme grant
Amount £2,528,786 (GBP)
Funding ID EP/R018472/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start  
 
Title NetEMD code 
Description This code is almost ready to be published; the corresponding publication is under revision. 
Type Of Material Computer model/algorithm 
Provided To Others? No  
Impact We are using the alpha version of the code for network comparison projects at the Alan Turing Institute. 
 
Title Netdis 
Description This is the code for the network comparison method Netdis which we published in W. Ali, T. Rito, G. Reinert, F. Sun, and C. M. Deane (2014) Alignment-free protein interaction network comparison. Bioinformatics (2014) 30, i430-i437 
Type Of Material Computer model/algorithm 
Year Produced 2015 
Provided To Others? Yes  
Impact Sarika Jalan's group at the ITT Indore is using Netdis to analyse networks arising from fMRI data. Drew Purves's group used it to compare ecological networks when he was at Microsoft Research. 
URL http://opig.stats.ox.ac.uk/resources
 
Title Subsampling code for Netdis 
Description This is the subsampling code for Netdis and other network comparison statistics which rely on local subgraph counts, as appeared in W. Ali, A. E. Wegner, R. E. Gaunt, C. M. Deane & G. Reinert (2016). Comparison of large networks with sub-sampling strategies. Scientific Reports 6: 28955. The algorithm is free to use but we have not yet been contacted by researchers who are interested in using it. 
Type Of Material Computer model/algorithm 
Provided To Others? No  
Impact We have used it to compare large networks. 
URL http://opig.stats.ox.ac.uk/resources
 
Description Alignment-free comparisons 
Organisation University of Southern California
Country United States 
Sector Academic/University 
PI Contribution We have contributed to the idea of Netdis and done the simulations; we also engage in alignment-free sequence comparison as a base-line for Netdis, and we have contributed to statistical estimation problems in this area. Moreover we continue to develop Stein's method as theoretical underpinning.
Collaborator Contribution Fruitful discussions regarding Netdis; concrete results on alignment-free sequence comparisons and related estimation problems.
Impact 1. W. Ali, T. Rito, G. Reinert, F. Sun, and C. M. Deane (2014) Alignment-free protein interaction network comparison. Bioinformatics (2014) 30, i430-i437. [Interdisciplinary: Mathematics and Biology] 2. L. Goldstein and G. Reinert (2013) Stein's method for the Beta distribution and the Pòlya-Eggenberger Urn. Journal of Applied Probability 1187-1205. 4. K. Song, J. Ren, G. Reinert, M. Deng, M. S. Waterman and F. Sun (2013) New developments of alignment-free sequence comparison: measures, statistics and next-generation sequencing. Briefings in Bioinformatics 10.1093/bib/bbt067. [Multi-disciplinary: Mathematics, Biology and Computer Science]
 
Description Copulas and networks 
Organisation Lloyd's Register
Country United Kingdom 
Sector Charity/Non Profit 
PI Contribution We provided the analysis and the model; and we are thinking about a network-related version of our work.
Collaborator Contribution The partner provided the research question and the data as well some input to the analysis.
Impact A two-component copula with links to insurance Samiha Ismail, Gao Yu, Gesine Reinert, Trevor Maynard Under revision; preprint available at http://arxiv.org/abs/1410.8740
Start Year 2012
 
Description Network time series comparison 
Organisation Alan Turing Institute
Country Unknown 
Sector Academic/University 
PI Contribution Supervision of a PDRA, joint with Dr Mihai Cucuringu, to use our network comparison methods for prediction in financial time series
Collaborator Contribution paying the PDRA (6 months)
Impact none yet; a paper in progress
Start Year 2017
 
Description Processes on networks 
Organisation University of Melbourne
Country Australia 
Sector Academic/University 
PI Contribution In this work on processes on networks, such as spread of information and disease, I have been providing the network expertise and some branching process expertise and contributed to the mathematical results.
Collaborator Contribution My partner has provided the branching process expertise and some network expertise and contributed to the mathematical results.
Impact A.D. Barbour and G. Reinert (2013) Asymptotic behaviour of gossip processes and small-world networks. Advances in Applied Probability 45, 895-1201.
 
Description Stein's method: the density approach 
Organisation University Libre Bruxelles (Université Libre de Bruxelles ULB)
Country Belgium 
Sector Academic/University 
PI Contribution We are providing expertise on Stein's method as well as research questions.
Collaborator Contribution They are providing a new angle on Stein's method. We are currently working on a multivariate version of Stein's method which can be applied to characterise distributions on the space of networks; we have a preprint (2017) with Dr Guillaume Mijoule and Prof. Yvik Swan.
Impact Ley, Christophe, Gesine Reinert, and Yvik Swan. "Stein's method for comparison of univariate distributions." Probability Surveys 14 (2017): 1-52. Ley, Christophe, Gesine Reinert, and Yvik Swan. "Distances between nested densities and a measure of the impact of the prior in Bayesian statistics." Annals of Applied Probability, in print.
Start Year 2012
 
Description Stein's method: the density approach 
Organisation University of Liege
Country Belgium 
Sector Academic/University 
PI Contribution We are providing expertise on Stein's method as well as research questions.
Collaborator Contribution They are providing a new angle on Stein's method. We are currently working on a multivariate version of Stein's method which can be applied to characterise distributions on the space of networks; we have a preprint (2017) with Dr Guillaume Mijoule and Prof. Yvik Swan.
Impact Ley, Christophe, Gesine Reinert, and Yvik Swan. "Stein's method for comparison of univariate distributions." Probability Surveys 14 (2017): 1-52. Ley, Christophe, Gesine Reinert, and Yvik Swan. "Distances between nested densities and a measure of the impact of the prior in Bayesian statistics." Annals of Applied Probability, in print.
Start Year 2012
 
Description Systemic risk on networks 
Organisation Technical University of Munich
Country Germany 
Sector Academic/University 
PI Contribution In the project on systemic risk on networks I have provided the network expertise and contributed to the detailed research questions as well as to the mathematical results.
Collaborator Contribution My partners have provided expertise on multivariate regular variation and contributed to the detailed research questions as well as to the mathematical results.
Impact O. Kley, C. Klüppelberg, and G. Reinert (2016). Risk in a Large Claims Insurance Market with Bipartite Graph Structure. Operations Research 64.5 (2016): 1159-1176. The paper provides a foundation for guidance to public authorities for managing insurance reserves; it discusses systemic Value-at-Risk and related risk measures. We were awarded a Science of Risk Prize by Lloyd's of London. Currently (2017) we are revising a paper on systemic risk measures in insurance markets, viewed as networks.
Start Year 2013
 
Description Threat detection in networks 
Organisation Accenture
Country Ireland 
Sector Private 
PI Contribution The Alan Turing Institute facilitated this collaboration on anomaly detection in networks. The methodology is based on the suite of network comparison methods which we developed. We are refining it and combining it with fast matrix methods.
Collaborator Contribution Accenture will provide expertise on behavioural analysis. They fund a postdoctoral research assistant for 2 years; the PDRA is based at the Alan Turing Institute.
Impact Still in progress.
Start Year 2017
 
Description UCB 
Organisation UCB Pharma
Country United Kingdom 
Sector Private 
PI Contribution Gesine Reinert and Charlotte Deane supervise a doctoral student, Malte Luecken, on using networks for macrophage differentiation, jointly with Matt Page, Andrea Crosby and Sean Mason from UCB.
Collaborator Contribution They provide biological expertise as well as funding; £9,000 to the University last year and £6,000 a year for the next 2 years, paid directly to the University.
Impact Transfer of status report: Malte Luecken: Application of Multi-Resolution Partitioning of Interaction Networks to the Study of Complex Disease Publications with UCB partners James Dunbar, Bernhard Knapp, Angelika Fuchs, Jiye Shi, Charlotte M Deane, Examining variable domain orientations in antigen receptors gives insight into TCR-like antibody design, Plos Comp. Bio, 2014, 10(9), e1003852 Henry R. Wilman, Jean-Paul Ebejer, Jiye Shi, Charlotte M Deane, Bernhard Knapp, Crowdsourcing yields a new standard for kinks in protein helices, Journal of Chemical Information and Modeling, 2014, 54(9), 2585-2593 K Krawczyk, X Liu, T Baker, J Shi, CM Deane, Improving B-cell epitope prediction and its application to global antibody-antigen docking, Bioinformatics, 2014, 30(16), 2288-2294 H.R. Wilman, J. Shi, C.M. Deane, Helix kinks are equally prevalent in soluble and membrane proteins, Proteins, 2014, 82(9), 1960-70
Start Year 2013
 
Description network comparison package 
Organisation Alan Turing Institute
Country Unknown 
Sector Academic/University 
PI Contribution We provided methods, expertise and vignettes as well as extensive testing.
Collaborator Contribution The Alan Turing Insittute provided funding for a month's time of a software engineer to turn our network comparison methods into an R package which satisfies professional standards.
Impact a GitHub package, https://www.turing.ac.uk/events/r-package-alignment-free-network-comparison/
Start Year 2017
 
Title NetEMD 
Description This is an R package implementing the NetEMD network comparison method which arose from the grant. 
Type Of Technology Webtool/Application 
Year Produced 2017 
Impact The package is still under development but will be finished by the end of March. It forms the basis of talks with Accenture about financial fraud, and also assists with a new project, joint with Niall Adams at Imperial, on cyber security. 
 
Title Netdis 
Description This stand-alone package calculates the Netdis distance. 
Type Of Technology Software 
Year Produced 2014 
Open Source License? Yes  
Impact This package disseminates the Netdis algorithm for network comparison. We have already had some queries about it. 
URL http://www.stats.ox.ac.uk/research/proteins/resources
 
Title Network comparison package 
Description This is an R package for our network comparison methods Netdis and NetEMD, professionally designed and with data sets and vignettes. We are working on a publication. 
Type Of Technology Software 
Year Produced 2017 
Open Source License? Yes  
Impact The software has been downloaded from external machines across Europe, and we have had some feedback. 
 
Title software for netdis 
Description This R package implements the subsampling appraoch for network comparison using the Netdis method. 
Type Of Technology Webtool/Application 
Year Produced 2016 
Impact The method is more widely used by other researchers. 
URL http://opig.stats.ox.ac.uk/resources
 
Description Alan Turing Institute scoping workshop 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact These events were two scoping meetings for the Alan Turing Institute, one on high-dimensional data analysis (September 28-30, Cambridge) and one on Networks and Big Data (December 18, London). Gesine Reinert gave strategic presentations on open questions in network analysis. The presentations sparked discussions with participants from industry (Barclay's Bank, Cray) as well as from other universities.
Year(s) Of Engagement Activity 2015
 
Description Blog post 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Public/other audiences
Results and Impact This is a blogpost on the popular opig blog website which describes our work on network comparison. This activity is ongoing. So far we have not been contacted with requests for more information but the visits to the opig blogs are monitored.
Year(s) Of Engagement Activity 2016
URL http://www.blopig.com/blog/
 
Description CDT lecture 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact This was a talk by Gesine Reinert on Stein's method with applications to networks questions.

After the talk a postdoctoral researcher approached me with a research question related to Markov chain Monte Carlo. After our discussion he was able to solve this question.
Year(s) Of Engagement Activity 2014
 
Description Chairing a discussion session 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Industry/Business
Results and Impact This was a workshop on ``The Age of Intelligence'' organised by the Accenture-Turing partnership. Jointly with Kieran Towey and Milton Martinez Luaces from Accenture I chaired a breakout session on AI in Fraud and Financial Crime. Participants came from banks, HMRC and other government agencies. .
Year(s) Of Engagement Activity 2018
URL https://s3.amazonaws.com/assets.accenture.com/The_Age_Of_Intelligence_Welcome_Pack_Turing.pdf
 
Description Complex 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? Yes
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact Gesine Reinert presented Netdis and there was considerable interest from MicroSoft Research.

Microsoft have requested our code and they are trying it on ecological network comparison.
Year(s) Of Engagement Activity 2014
URL http://www.ucl.ac.uk/complex/conference2014
 
Description DTC lectures 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact This was a series of lectures and practical exercises, spanning three days, given by Gesine Reinert, on network analysis methods and biological networks.

A few students contacted me to discuss with them whether their research question could be studied from a networks angle.
Year(s) Of Engagement Activity 2014
 
Description ECCB 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Our paper on Netdis was selected for a talk at the European Conference for Computational Biology in Strasbourg. Dr Waqar Ali gave the presentation and engaged the audience. This presentation also helped him as an early-career researcher to gain recognition.

This presentation helped Dr Ali who is an early-career researcher to gain recognition.
Year(s) Of Engagement Activity 2014
URL http://www.eccb14.org/programme/proceedings_papers_abstracts.pdf
 
Description Heriot Watt 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Seminar talk by Dr Robert Gaunt: Rates of convergence for multivariate normal approximations by Stein's method

The talk is part of a visit by Robert Gaunt to Fraser Daly at Heriot Watt, with the purpose to stimulate collaboration and to raise Dr Gaunt's professional visibility.
Year(s) Of Engagement Activity 2014
 
Description IWAP 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Gesine Reinert gave an invited lecture on Stein's method with applications.

After the talk there was considerable discussion about application to Bayesian analysis.
Year(s) Of Engagement Activity 2014
URL http://www.iwap2014.org/?p=invited_sessions
 
Description IWAP Gaunt 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talk generated discussion afterwards.

After the talk there was some discussion.
Year(s) Of Engagement Activity 2014
URL http://www.iwap2014.org/?p=Contributed_Sessions
 
Description Limit Theorems 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact An invited talk (one of the three speakers) at the ``Workshop on Limit theorems, probability approximations and related areas''

After the talk there was discussion about relationships to epidemics on networks (with Denis Mollison)
Year(s) Of Engagement Activity 2014
URL https://www.google.com/calendar/event?eid=Y3BkYnJrYTR0azA1b2w3M2Q4NG1vamg3NzggYzE2bG5vYm8xYXZ0M3ZvYW...
 
Description Luminy 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? Yes
Type Of Presentation workshop facilitator
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact The workshop combined different approaches to limit theorems, bringing together researchers from different areas of mathematics and statistics.

After the workshop a number of new collaborations arose.
Year(s) Of Engagement Activity 2014
 
Description Missilac 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? Yes
Geographic Reach International
Primary Audience Other academic audiences (collaborators, peers etc.)
Results and Impact Talk on Stein's method and applications; with plenty of discussion afterwards

A new research collaboration on concentration inequalities arose.
Year(s) Of Engagement Activity 2014
URL http://dornsife.usc.edu/conferences/steincolloquium/
 
Description Outreach talk 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Schools
Results and Impact Talk at the UNIQ Summer School in Oxford

The students were keen and interested in software for further study
Year(s) Of Engagement Activity 2015
 
Description Participation in workshop with industry 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact LLoyd's Science of Risk workshop (invitation only, most participants were insurance professionals across a range of businesses)

I gave a talk about systemic risk in networks. With my co-authors Oliver Kley and Claudia Klueppelberg we were presented with a Science of Risk Prize.

There will be a follow-up meeting later this month.
Year(s) Of Engagement Activity 2016
 
Description Royal Statistical Society workshop on networks: organisation 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact This was a workshop at the RSS which the PI co-organised; it reached out to computer scientists, life scientists and physicists as well as being of interest to probabilists and statisticians. There was some participation from industry as well.
Year(s) Of Engagement Activity 2016
URL https://www.statslife.org.uk/events/eventdetail/572/-/networks
 
Description Talk at Oxford Prospects Programme 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Undergraduate students
Results and Impact This was a talk on network analysis at the Oxford Prospects Programmes (OPP) . The OPP is a collaboration sponsored by Regent's Park and Shanghai MEC which works to promote education exchanges between institutions of higher education in China and the UK. www.oxford-prospects.org
Year(s) Of Engagement Activity 2017
URL http://www.rpc.ox.ac.uk/opgdc/oxford-prospects-programmes/
 
Description Talk at Oxford Prospects meeting 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Undergraduate students
Results and Impact An introductory talk on network analysis for Chinese students who are interested in postgraduate studies in the UK
Year(s) Of Engagement Activity 2019
 
Description Turing Strategic Partners Board - highlight presentation 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact October 27, 2017: Highlight talk on network analysis to the Turing Strategic Partners Board (SPB) . At this meeting, top executives from Lloyd's Register Foundation, Intel, HSBC and GCHQ+ Dstl were attending. The SPB's role is to advice the Turing Board of Trustees on partners' priorities and also to discuss progress with individual partners' programmes and cross-over between these (and opportunities for further collaboration).
Year(s) Of Engagement Activity 2017
 
Description Turing data science class 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact This was a 2-hour lecture course on network analysis, held at the Turing and broadcast nationwide.
Year(s) Of Engagement Activity 2018
URL https://www.turing.ac.uk/events/statistical-analysis-networks-part-12/
 
Description UNIQ summer school talk 2015, 2016 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Schools
Results and Impact a one-hour talk on networks, given to A level pupils, as part of the UNIQ Mathematics summer school which happen in Oxford. In 2015 I gave the talk, in 2016 Robert Gaunt gave my talk.
Year(s) Of Engagement Activity 2015,2016
 
Description Workshop with industry 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Industry/Business
Results and Impact This was a 1-day workshop on network pharmacology, which brought together participants from industry (e-Therapeutics, Genomics plc), graduate students, postdoctoral researchers and faculty
Year(s) Of Engagement Activity 2015
URL http://www.keble-asc.com/?id=1547