Inference, COmputation and Numerics for Insights into Cities (ICONIC)
Lead Research Organisation:
Imperial College London
Department Name: Mathematics
Abstract
There are many interesting open questions at the interface between applied mathematics, scientific computing and applied statistics.
Mathematics is the language of science, we use it to describe the laws of motion that govern natural and technological
systems. We use statistics to make sense of data. We develop and test computer algorithms that make these ideas concrete. By bringing these concepts together in a systematic way we can validate and sharpen our hypothesis about the underlying science, and make predictions about future behaviour. This general field of Uncertainty Quantification is a very active area of research, with many challenges; from intellectual questions about how to define and measure uncertainty to very practical issues concerning the need to perform intensive computational experiments as efficiently as possible.
ICONIC brings together a team of high profile researchers with the appropriate combination of skills in modeling, numerical analysis, statistics and high performance computing. To give a concrete target for impact, the ICONIC project will focus initially on Uncertainty Quantification for mathematical models relating to crime, security and resilience in urban environments. Then, acknowledging that urban analytics is a very fast-moving field where new technologies and data sources emerge rapidly, and exploiting the flexibility built into an EPSRC programme grant, we will apply the new tools to related city topics concerning human mobility, transport and infrastructure. In this way, the project will enhance the UK's research capabilities in the fast-moving and globally significant Future Cities field.
The project will exploit the team's strong existing contacts with Future Cities laboratories around the world, and with nonacademic stakeholders who are keen to exploit the outcomes of the research. As new technologies emerge, and as more people around the world choose to live and work in urban environments, the Future Cities field is generating vast quantities of potentially valuable data. ICONIC will build on the UK's strength in basic mathematical sciences--the cleverness needed to add value to these data sources--in order to produce new algorithms and computational tools. The research will be conducted alongside stakeholders--including law enforcement agencies, technical IT and infrastructure providers, utility companies and policy-makers. These external partners will provide feedback and challenges, and will be ready to extract value from the tools that we develop. We also have an international Advisory Board of committed partners with relevant expertise in academic research, policymaking, law enforcement, business engagement and public outreach. With these structures in place, the research will have a direct impact on the UK economy, as the nation competes for business in the global Future Cities marketplace. Further, by focusing on crime, security and resilience we will directly improve the lives of individual citizens.
Mathematics is the language of science, we use it to describe the laws of motion that govern natural and technological
systems. We use statistics to make sense of data. We develop and test computer algorithms that make these ideas concrete. By bringing these concepts together in a systematic way we can validate and sharpen our hypothesis about the underlying science, and make predictions about future behaviour. This general field of Uncertainty Quantification is a very active area of research, with many challenges; from intellectual questions about how to define and measure uncertainty to very practical issues concerning the need to perform intensive computational experiments as efficiently as possible.
ICONIC brings together a team of high profile researchers with the appropriate combination of skills in modeling, numerical analysis, statistics and high performance computing. To give a concrete target for impact, the ICONIC project will focus initially on Uncertainty Quantification for mathematical models relating to crime, security and resilience in urban environments. Then, acknowledging that urban analytics is a very fast-moving field where new technologies and data sources emerge rapidly, and exploiting the flexibility built into an EPSRC programme grant, we will apply the new tools to related city topics concerning human mobility, transport and infrastructure. In this way, the project will enhance the UK's research capabilities in the fast-moving and globally significant Future Cities field.
The project will exploit the team's strong existing contacts with Future Cities laboratories around the world, and with nonacademic stakeholders who are keen to exploit the outcomes of the research. As new technologies emerge, and as more people around the world choose to live and work in urban environments, the Future Cities field is generating vast quantities of potentially valuable data. ICONIC will build on the UK's strength in basic mathematical sciences--the cleverness needed to add value to these data sources--in order to produce new algorithms and computational tools. The research will be conducted alongside stakeholders--including law enforcement agencies, technical IT and infrastructure providers, utility companies and policy-makers. These external partners will provide feedback and challenges, and will be ready to extract value from the tools that we develop. We also have an international Advisory Board of committed partners with relevant expertise in academic research, policymaking, law enforcement, business engagement and public outreach. With these structures in place, the research will have a direct impact on the UK economy, as the nation competes for business in the global Future Cities marketplace. Further, by focusing on crime, security and resilience we will directly improve the lives of individual citizens.
Planned Impact
The advances in fundamental research arising from the programme will impact academic researchers in applied/computational
mathematics, statistics and computer science; in particular those working in numerical analysis, statistical inference,
uncertainty quantification, machine learning, data science, network science, stochastic simulation, mathematical modelling, data assimilation and high performance computing. Moreover, the application-oriented results will impact colleagues in geography, urban design, architecture, transport, crime studies, politics and policy making.
Outside academia, because the project is embedded within a Future Cities framework, with emphasis on security, crime and resilience, immediate beneficiaries will include
- SMEs who develop bespoke data analytics applications for clients in local and national government, who will gain access to more insightful data analytics tools that can be customized to their needs,
- larger hi-tech companies developing complex IT solutions across multiple platforms,
- clients of these companies, including those in transport, sensors, satellite imaging, construction, energy and other
utilities, charities and governments, who will benefit from the extra efficincies and insights through improved products and services,
- companies working in tourism, event planning, sports and retail who will benefit from quantified city-level information,
- policy makers and law enforcement officers, who will be better placed to make informed decisions,
- user-communities living, working or spending leisure time in a city environment, who will experience better and safer
services.
A key aspect of the impact delivery plan is to exploit our existing, strong links with important players at the
academic/external interface. In particular, we have in-house connections with Imperial's Digital City Exchange, the Oxford Martin School's programme in Cities, in Future Technology, and in Science and Society, and Strathclyde's Institute for Future Cities. We also have very close working relationships with the UK's Future Cities Catapult, Dublin's Trinity Centre for Smart and Sustainable Cities, New York's Center for Urban Science and Progress (CUSP) and Chicago's Urban Centre for Computation and Data. These partners give an excellent mechanism for outreach to relevant stakeholders and for public engagement, with a level and scope that would not be possible from a single, isolated programme.
Girolami and D. Higham have been drivers in the ATI continuing development of an Urban Analytics research theme. ICONIC will be completely distinct from this and fully complementary, it is driven by the urgent need for fundamental research in mathematical sciences, driven by mathematical and statistical modelling challenges that form an holistic UQ pipeline. ATI is viewed as a UK organisation that can exploit and translate the novel tools emerging from ICONIC and co-engage in additional outreach and impact generation. Similarly, D. Higham's Digital Economy Fellowship, which employs the PDRA Dr Francesca Arrigo as a Data Scientist and has a focus on dynamic networks of digital interactions, has external partners such as Bloom Agency, Capita, CountingLab & Siemens, who can make rapid use of ICONIC's research outputs.
Relevant new links to external partners have also been established: see letters of support from representatives of UK Government Home Office, Metropolitan Police Service, New York Police Department, Police Scotland & West Midlands Police. In addition to shaping and stress-testing the research, these partners will also provide routes to rapid and high-value deployment.
Further, by training PDRAs, the project will deliver skilled, outward facing, future-leaders with experience of interacting with external partners. These individuals will be well placed to act as agents-for-change in shaping the UK demand for Future City Analytics in way that strengthens UK competitiveness.
mathematics, statistics and computer science; in particular those working in numerical analysis, statistical inference,
uncertainty quantification, machine learning, data science, network science, stochastic simulation, mathematical modelling, data assimilation and high performance computing. Moreover, the application-oriented results will impact colleagues in geography, urban design, architecture, transport, crime studies, politics and policy making.
Outside academia, because the project is embedded within a Future Cities framework, with emphasis on security, crime and resilience, immediate beneficiaries will include
- SMEs who develop bespoke data analytics applications for clients in local and national government, who will gain access to more insightful data analytics tools that can be customized to their needs,
- larger hi-tech companies developing complex IT solutions across multiple platforms,
- clients of these companies, including those in transport, sensors, satellite imaging, construction, energy and other
utilities, charities and governments, who will benefit from the extra efficincies and insights through improved products and services,
- companies working in tourism, event planning, sports and retail who will benefit from quantified city-level information,
- policy makers and law enforcement officers, who will be better placed to make informed decisions,
- user-communities living, working or spending leisure time in a city environment, who will experience better and safer
services.
A key aspect of the impact delivery plan is to exploit our existing, strong links with important players at the
academic/external interface. In particular, we have in-house connections with Imperial's Digital City Exchange, the Oxford Martin School's programme in Cities, in Future Technology, and in Science and Society, and Strathclyde's Institute for Future Cities. We also have very close working relationships with the UK's Future Cities Catapult, Dublin's Trinity Centre for Smart and Sustainable Cities, New York's Center for Urban Science and Progress (CUSP) and Chicago's Urban Centre for Computation and Data. These partners give an excellent mechanism for outreach to relevant stakeholders and for public engagement, with a level and scope that would not be possible from a single, isolated programme.
Girolami and D. Higham have been drivers in the ATI continuing development of an Urban Analytics research theme. ICONIC will be completely distinct from this and fully complementary, it is driven by the urgent need for fundamental research in mathematical sciences, driven by mathematical and statistical modelling challenges that form an holistic UQ pipeline. ATI is viewed as a UK organisation that can exploit and translate the novel tools emerging from ICONIC and co-engage in additional outreach and impact generation. Similarly, D. Higham's Digital Economy Fellowship, which employs the PDRA Dr Francesca Arrigo as a Data Scientist and has a focus on dynamic networks of digital interactions, has external partners such as Bloom Agency, Capita, CountingLab & Siemens, who can make rapid use of ICONIC's research outputs.
Relevant new links to external partners have also been established: see letters of support from representatives of UK Government Home Office, Metropolitan Police Service, New York Police Department, Police Scotland & West Midlands Police. In addition to shaping and stress-testing the research, these partners will also provide routes to rapid and high-value deployment.
Further, by training PDRAs, the project will deliver skilled, outward facing, future-leaders with experience of interacting with external partners. These individuals will be well placed to act as agents-for-change in shaping the UK demand for Future City Analytics in way that strengthens UK competitiveness.
Organisations
- Imperial College London (Lead Research Organisation)
- West Midlands Police (Project Partner)
- New York Police Department (Project Partner)
- MathWorks (United Kingdom) (Project Partner)
- Smith Institute (Project Partner)
- Metropolitan Police Service (Project Partner)
- Police Scotland (Project Partner)
- New York University (Project Partner)
- Future Cities Catapult (United Kingdom) (Project Partner)
Publications
Giles M
(2023)
Approximating Inverse Cumulative Distribution Functions to Produce Approximate Random Variables
in ACM Transactions on Mathematical Software
Amestoy P
(2023)
Combining Sparse Approximate Factorizations with Mixed-precision Iterative Refinement
in ACM Transactions on Mathematical Software
Fasi M
(2023)
CPFloat: A C Library for Simulating Low-precision Arithmetic
in ACM Transactions on Mathematical Software
Higham N
(2022)
Mixed precision algorithms in numerical linear algebra
in Acta Numerica
Virtanen S.
(2019)
Precision-recall balanced topic modelling
in Advances in Neural Information Processing Systems
Barp A
(2018)
Geometry and Dynamics for Markov Chain Monte Carlo
in Annual Review of Statistics and Its Application
Tudisco F
(2019)
A fast and robust kernel optimization method for core-periphery detection in directed and weighted graphs
in Applied Network Science
Cockayne J
(2019)
A Bayesian Conjugate Gradient Method (with Discussion)
in Bayesian Analysis
Livingstone S
(2019)
On the geometric ergodicity of Hamiltonian Monte Carlo
in Bernoulli
Oates C
(2019)
Convergence rates for a class of estimators based on Stein's method
in Bernoulli
Barp A
(2022)
A Riemann-Stein kernel method
in Bernoulli
John Higham D
(2022)
Disease extinction for susceptible-infected-susceptible models on dynamic graphs and hypergraphs.
in Chaos (Woodbury, N.Y.)
Higham DJ
(2020)
A network model for polarization of political opinion.
in Chaos (Woodbury, N.Y.)
Tudisco F
(2021)
Node and edge nonlinear eigenvector centrality for hypergraphs
in Communications Physics
Girolami M
(2021)
The statistical finite element method (statFEM) for coherent synthesis of observation data and model predictions
in Computer Methods in Applied Mechanics and Engineering
Anzt H
(2018)
Adaptive precision in block-Jacobi preconditioning for iterative sparse linear system solvers
in Concurrency and Computation: Practice and Experience
De Kergorlay H
(2023)
Connectivity of Random Geometric Hypergraphs
in Entropy
GILMOUR C
(2021)
Modelling Burglary in Chicago using a self-exciting point process with isotropic triggering
in European Journal of Applied Mathematics
Fasi M
(2021)
Algorithms for Stochastically Rounded Elementary Arithmetic Operations in IEEE 754 Floating-Point Arithmetic
in IEEE Transactions on Emerging Topics in Computing
Tyukin I
(2023)
The feasibility and inevitability of stealth attacks
in IMA Journal of Applied Mathematics
Beerens L
(2023)
Adversarial ink: componentwise backward error attacks on deep learning
in IMA Journal of Applied Mathematics
Blanchard P
(2021)
Accurately computing the log-sum-exp and softmax functions
in IMA Journal of Numerical Analysis
Croci M
(2023)
Effects of round-to-nearest and stochastic rounding in the numerical solution of the heat equation in low precision
in IMA Journal of Numerical Analysis
Higham N
(2022)
Solving block low-rank linear systems by LU factorization is numerically stable
in IMA Journal of Numerical Analysis
De Kergorlay H
(2022)
Consistency of anchor-based spectral clustering
in Information and Inference: A Journal of the IMA
Klöwer M
(2022)
Fluid Simulations Accelerated With 16 Bits: Approaching 4x Speedup on A64FX by Squeezing ShallowWaters.jl Into Float16
in Journal of Advances in Modeling Earth Systems
Manderson A
(2019)
Uncertainty Quantification of Density and Stratification Estimates with Implications for Predicting Ocean Dynamics
in Journal of Atmospheric and Oceanic Technology
Higham D
(2019)
Centrality-friendship paradoxes: when our friends are more important than us
in Journal of Complex Networks
Giles M
(2019)
Random bit multilevel algorithms for stochastic differential equations
in Journal of Complexity
Gregory A
(2019)
The synthesis of data from instrumented structures and physics-based models via Gaussian processes
in Journal of Computational Physics
Dunlop Matthew M.
(2018)
How Deep Are Deep Gaussian Processes?
in JOURNAL OF MACHINE LEARNING RESEARCH
Fang W
(2019)
Multilevel Monte Carlo method for ergodic SDEs without contractivity
in Journal of Mathematical Analysis and Applications
Arrigo F
(2019)
Non-backtracking PageRank
in Journal of Scientific Computing
Oates C
(2019)
Bayesian Probabilistic Numerical Methods in Time-Dependent State Estimation for Industrial Hydrocyclone Equipment
in Journal of the American Statistical Association
Virtanen S
(2021)
Spatio-Temporal Mixed Membership Models for Criminal Activity
in Journal of the Royal Statistical Society Series A: Statistics in Society
Povala J
(2020)
Burglary in London: Insights from Statistical Heterogeneous Spatial Point Processes
in Journal of the Royal Statistical Society Series C: Applied Statistics
Pranesh S
(2020)
Backward error and condition number of a generalized Sylvester equation, with application to the stochastic Galerkin method
in Linear Algebra and its Applications
Higham N
(2021)
Integer matrix factorisations, superalgebras and the quadratic form obstruction
in Linear Algebra and its Applications
Arrigo F
(2022)
Dynamic Katz and related network measures
in Linear Algebra and its Applications
Fang W
(2022)
Multilevel and Quasi Monte Carlo Methods for the Calculation of the Expected Value of Partial Perfect Information.
in Medical decision making : an international journal of the Society for Medical Decision Making
Di Giovacchino S
(2022)
A Hierarchy of Network Models Giving Bistability Under Triadic Closure
in Multiscale Modeling & Simulation
Anderson D
(2019)
On Constrained Langevin Equations and (Bio)Chemical Reaction Networks
in Multiscale Modeling & Simulation
Higham N
(2021)
Anymatrix: an extensible MATLAB matrix collection
in Numerical Algorithms
Glyn-Davies A
(2022)
Anomaly detection in streaming data with gaussian process based stochastic differential equations
in Pattern Recognition Letters
Zounon M
(2022)
Performance impact of precision reduction in sparse linear systems solvers
in PeerJ Computer Science
Fasi M
(2021)
Numerical behavior of NVIDIA tensor cores.
in PeerJ. Computer science
Duffin C
(2021)
Statistical finite elements for misspecified models.
in Proceedings of the National Academy of Sciences of the United States of America
Arrigo F
(2020)
A framework for second-order eigenvector centralities and clustering coefficients.
in Proceedings. Mathematical, physical, and engineering sciences
Higham DJ
(2021)
Epidemics on hypergraphs: spectral thresholds for extinction.
in Proceedings. Mathematical, physical, and engineering sciences
Ellam L
(2018)
Stochastic modelling of urban structure.
in Proceedings. Mathematical, physical, and engineering sciences
Description | Mathematics of Adversarial Attacks |
Amount | £202,126 (GBP) |
Funding ID | EP/V046527/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 01/2021 |
End | 12/2022 |
Title | MATLAB Software for Deep Learning |
Description | Scientific Computing software to accompany an expository article on the mathematics of deep learning |
Type Of Material | Improvements to research infrastructure |
Year Produced | 2018 |
Provided To Others? | Yes |
Impact | Accompanying expository article has been used in several universities as the basis for classes, tutorials and reading groups on deep learning. |
URL | http://personal.strath.ac.uk/d.j.higham/algfiles.html |
Description | LMS/IMA |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | Invited talk at London Mathematical Society/Society for Industrial and Applied Mathematics joint meeting on 30th September and 1st October. Hosted by the ICMS (Edinburgh), addressing the theme of 'Mathematics in Human Society'. |
Year(s) Of Engagement Activity | 2021 |
URL | https://ima.org.uk/17272/lms-ima-joint-meeting-2021-maths-in-human-society/ |
Description | Pint of Science talk in Glasgow, 22 May 2019 |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Regional |
Primary Audience | Public/other audiences |
Results and Impact | Pint of Science is a regular public engagement initiative, with selected science speakers. |
Year(s) Of Engagement Activity | 2019 |
URL | https://pintofscience.co.uk/city/glasgow |
Description | Research talk at Skolkovo |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | Invited research talk and round table panel membership at Trustworthy AI 5-7 July Skolkovo, Moscow My attendance was virtual. |
Year(s) Of Engagement Activity | 2012,2021 |
URL | https://events.skoltech.ru/ai-trustworthy#content |
Description | Turing mtg |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | Invited research talk at Interpretability, safety and security in AI conference 13-15th December, Alan Turing Institute (virtual attendance) |
Year(s) Of Engagement Activity | 2021 |
URL | https://www.turing.ac.uk/events/interpretability-safety-and-security-ai |
Description | Workshop on Data Science and Crime |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | One day workshop involving police data scientists from West Midlands, Police Scotland and ondon Met, and academics with relevant skills. Knowledge exchange and highlighting of current challenges were the key aims. |
Year(s) Of Engagement Activity | 2018 |
URL | https://iconicmath.org/ds-crime-workshop/ |
Description | Workshop on Data Science and Soaicl Media |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Industry/Business |
Results and Impact | An afternoon Workshop on Data Science and Social Media, held as part of a wekk ling Engage with Strathclyde series of events. About 30 people attended from social media, policymaking, economics, to share best practice and learn about new developments at the intersection of Data Science and Social media. |
Year(s) Of Engagement Activity | 2018 |
URL | https://www.engage.strath.ac.uk/ |
Description | faculty lecture |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Postgraduate students |
Results and Impact | Invited faculty lecture (online) Deep Learning: what could go wrong April 2021 |
Year(s) Of Engagement Activity | 2020,2021 |
URL | https://www.youtube.com/watch?v=yVXtoizLl8U |
Description | research worksop |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | Dagstuhl Seminar on 'Higher-Order Graph Models: From Theoretical Foundations to Machine Learning' (21352) August 29- Sep 1, 2021 |
Year(s) Of Engagement Activity | 2021 |
URL | https://www.dagstuhl.de/en/program/calendar/semhp/?semnr=21352 |