Dr Steven Reece

Head of Machine Learning Research and Data Science, Oxford Sustainable Finance Group

Profile

Steve is head of the Machine Learning Research and Data Science cluster in the Oxford Sustainable Finance Group. His team exploits and further develops state-of-the-art machine learning for mapping and characterising assets world-wide that exhibit both environmental and financial risk. Current work is focussed on creating asset level datasets of the world's most polluting industries using remote sensing and unstructured text reports.

Prior to joining SSEE in November 2021, Steve was simultaneously a Departmental Research Fellow in machine learning in the Engineering Science Department, Oxford University, a Turing Fellow at the Alan Turing Institute and an EPSRC funded Researcher in Residence at the Satellite Applications Catapult. He led a research group in disaster response and environmental protection, working at the intersection of Bayesian and deep learning for scalable human-agent computation, intelligent crowd-sourced data labelling, interpretable machine learning, concept formation and heterogeneous data fusion.

He has over 25 years research and consultancy experience in machine learning, over 70 publications in the field and also patents in defence and commercial applications. His methods have been applied to a diverse range of problems including fault correction of NASA astrophysics data, multi-robot terrain mapping and anomaly detection, social network analysis, sensor modelling and big data economics. Steve has collaborated closely with disaster response organisations and has developed and deployed satellite imagery analysis tools to provide the UN, FEMA and NGOs with damage situation maps immediately following natural disasters. His recent collaborations have been with hydrologists here in the School of Geography on rainfall-runoff modelling in the UK, with hydro-meteorological commercial organisations on flood impact in India, with commercial maritime disaster prevention organisations on pollution source identification, with the Brazilian National Audit Office on tailings dam disaster prevention and illegal mining and with Save the Elephants on human-animal conflict.

Steve received his DPhil from the Department of Engineering Science, Oxford University, the Diploma in Computer Science and Part III of the Mathematics Tripos from Cambridge University and a BSc in Physics and Mathematics from the University of Liverpool.

Project

Machine learning & data science

Publications

Journal (44)

Open

Tkachenko N, Tang K, McCarten M, and others, ‘Global Database of Cement Production Assets and Upstream Suppliers’ (2023) 10 Scientific Data

Lees T, Reece S, Kratzert F, and others, ‘Hydrological Concept Formation Inside Long Short-Term Memory (LSTM) Networks’ (2022) 26 Hydrology and Earth System Sciences 3079

Lees T, Reece S, Kratzert F, and others, ‘Hydrological Concept Formation Inside Long Short-Term Memory (LSTM) Networks’ (2022) 26 HYDROLOGY AND EARTH SYSTEM SCIENCES 3079

Simmons B, Lintott C, Reece S, and others, ‘Disaster, Infrastructure and Participatory Knowledge: The Planetary Response Network’ (2022) 7 Citizen Science: Theory and Practice 21

Lees T, Tseng G, Atzberger C, and others, ‘Deep Learning for Vegetation Health Forecasting: A Case Study in Kenya’ (2022) 14 Remote Sensing

Lees T, Buechel M, Anderson B, and others, ‘Benchmarking Data-Driven Rainfall-Runoff Models in Great Britain: A Comparison of LSTM-Based Models With Four Lumped Conceptual Models’ (2021) 25 Hydrology and Earth System Sciences 5517

Lees T, Buechel M, Anderson B, and others, ‘Benchmarking Data-Driven Rainfall-Runoff Models in Great Britain: A Comparison of Long Short-Term Memory (LSTM)-Based Models With Four Lumped Conceptual Models’ (2021) 25 HYDROLOGY AND EARTH SYSTEM SCIENCES 5517

Duporge I, Isupova O, Reece S, and others, ‘Using Very-High-Resolution Satellite Imagery and Deep Learning to Detect and Count African Elephants in Heterogeneous Landscapes’ (2020) 7 Remote Sensing in Ecology and Conservation 369

Balaniuk R, Isupova O and Reece S, ‘Mining and Tailings Dam Detection in Satellite Imagery Using Deep Learning’ (2020) 20 Sensors

Aigrain S, Parviainen H, Roberts S, and others, ‘Robust, Open-Source Removal of Systematics in Kepler Data’ (2017) 471 Monthly Notices of the Royal Astronomical Society 759

Ramchurn S, Huynh T, Wu F, and others, ‘A Disaster Response System Based on Human-Agent Collectives’ (2016) 57 Journal of Artificial Intelligence Research 661

Ramchurn S, Wu F, Jiang W, and others, ‘Human–agent Collaboration for Disaster Response’ (2016) 30 Autonomous Agents and Multi-Agent Systems 82

Rajpaul V, Aigrain S, Osborne M, and others, ‘A Gaussian Process Framework for Modelling Stellar Activity Signals in Radial Velocity Data’ (2015) 452 Monthly Notices of the Royal Astronomical Society 2269

Ghosh S, Reece S, Rogers A, and others, ‘Modeling the Thermal Dynamics of Buildings: A Latent-Force- Model-Based Approach’ [2015] ACM Transactions on Intelligent Systems and Technology

Reece S, Ghosh S, Rogers A, and others, ‘Efficient State-Space Inference of Periodic Latent Force Models’ (2014) 15 Journal of Machine Learning Research 2337

Smith M, Reece S, Roberts S, and others, ‘Maritime Abnormality Detection Using Gaussian Processes’ (2014) 38 Knowledge and Information Systems 717

Roberts S, McQuillan A, Reece S, and others, ‘Astrophysically Robust Systematics Removal Using Variational Inference: Application to the First Month of Kepler Data’ (2013) 435 Monthly Notices of the Royal Astronomical Society 3639

Fischer J, Ramchurn S, Osborne M, and others, ‘Recommending Energy Tariffs and Load Shifting Based on Smart Household Usage Profiling’ 383

Roberts S, Osborne M, Ebden M, and others, ‘Gaussian Processes for Time-Series Modelling.’ (2013) 371 Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 20110550

Smith M, Reece S, Roberts S, and others, ‘Online Maritime Abnormality Detection Using Gaussian Processes and Extreme Value Theory’ 645

Reece S, Roberts S, Nicholson D, and others, ‘Determining Intent Using Hard Soft Data and Gaussian Process Classifiers’ [2011] Fusion 2011 - 14th International Conference on Information Fusion

Reece S, Mann R, Rezek I, and others, ‘Gaussian Process Segmentation of Co‐Moving Animals’ (2011) 1305 AIP Conference Proceedings 430

Garnett R, Osborne M, Reece S, and others, ‘Sequential Bayesian Prediction in the Presence of Changepoints and Faults’ (2010) 53 Computer Journal 1430

Reece S and Roberts S, ‘The Near Constant Acceleration Gaussian Process Kernel for Tracking’ (2010) 17 IEEE Signal Processing Letters 707

Reece S and Roberts S, ‘An Introduction to Gaussian Processes for the Kalman Filter Expert’ 1

Reece S and Roberts S, ‘Generalised Covariance Union: A Unified Approach to Hypothesis Merging in Tracking’ (2010) 46 IEEE Transactions on Aerospace and Electronic Systems 207

Reece S, Roberts S, Claxton C, and others, ‘Multi-Sensor Fault Recovery in the Presence of Known and Unknown Fault Types’ [2009] 2009 12th International Conference on Information Fusion, FUSION 2009 1695

Rezek I, Leslie D, Reece S, and others, ‘On Similarities Between Inference in Game Theory and Machine Learning’ (2008) 33 Journal of Artificial Intelligence Research 259

Reece S, Roberts S, Rogers A, and others, ‘A Multi-Dimensional Trust Model for Heterogeneous Contract Observations’ (2007) 1 Proceedings of the National Conference on Artificial Intelligence 128

Proudler I, Roberts S, Reece S, and others, ‘An Iterative Signal Detection Algorithm Based On Bayesian Belief Propagation Ideas’ 355

Reece S, Rogers A, Roberts S, and others, ‘Rumours and Reputation’ 1

Rogers A, Dash R, Jennings N, and others, ‘Computational Mechanism Design for Information Fusion Within Sensor Networks’ 1

Rogers A, Dash R, Jennings N, and others, ‘Computational Mechanism Design for Multi-Sensor Information Fusion’ 1463

Reece S and Nicholson D, ‘Tighter Alternatives to the Cramér-Rao Lower Bound for Discrete-Time Filtering’ 1

Reece S and Roberts S, ‘Robust, Low-Bandwidth, Multi-Vehicle Mapping’ 1319

Dash R, Rogers A, Jennings N, and others, ‘Constrained Bandwidth Allocation in Multi-Sensor Information Fusion: A Mechanism Design Approach’ 1185

Rolls E, Franco L, Aggelopoulos N, and others, ‘An Information Theoretic Approach to the Contributions of the Firing Rates and the Correlations Between the Firing of Neurons.’ (2003) 89 Journal of neurophysiology 2810

Reece S, Simmons C, Fitzhenry R, and others, ‘Mutagenesis of Conserved Tryptophan Residues Within the Receptor-Binding Domain of Intimin: Influence on Binding Activity and Virulence’ (2002) 148 Microbiology 657

Reece S, ‘Erratum: Nonlinear Kalman Filtering With Semi-Parametric Biscay Distributions (IEEE Transactions on Signal Processing (Nov. 2001) 49 (2445-2453))’ (2002) 50 IEEE Transactions on Signal Processing 990

Reece S, ‘Nonlinear Kalman Filtering With Semi-Parametric Biscay Distributions (vol 49, Pg 2445, 2001)’ (2002) 50 IEEE TRANSACTIONS ON SIGNAL PROCESSING 990

Reece S, ‘Nonlinear Kalman Filtering With Semi-Parametric Biscay Distributions’ (2001) 49 IEEE TRANSACTIONS ON SIGNAL PROCESSING 2445

Price E, Reece S and Probert-Smith P, ‘Statistical Signatures for Self-Adaptive Sensing’ [2001] MFI2001: INTERNATIONAL CONFERENCE ON MULTISENSOR FUSION AND INTEGRATION FOR INTELLIGENT SYSTEMS 141

Cawsey A, Galliers J, Reece S, and others, ‘Automating the Librarian: Belief Revision As a Base for System Action and Communication With the User’ (1992) 35 Computer Journal 221

Chapter (2)

Open

Nicholson D, Reece S, Rogers A, and others, ‘Distributed Data Fusion Overarching Design Concerns and Some New Approaches’, Distributed Data Fusion for Network-Centric Operations (2017)

Reece S and Nicholson D, ‘Tighter Alternatives to the cramér-Rao Lower Bound for Discretetime Filtering’, Bayesian Bounds for Parameter Estimation and Nonlinear Filtering/Tracking (2007)

Conference paper (16)

Open

Rossi C, Tkachenko N, Bayaraa M, and others, ‘Detection and Characterisation of Pollutant Assets With AI and EO to Prioritise Green Investments: The Geoasset Framework’ (Institute of Electrical and Electronics Engineers (IEEE) 2022)

Simpson E, Reece S and Roberts S, ‘Bayesian Heatmaps: Probabilistic Classification With Multiple Unreliable Information Sources’, ECML PKDD 2017: Machine Learning and Knowledge Discovery in Databases (Springer, Cham 2017)

Howard C, Jones D, Reece S, and others, ‘Learning to Trust the Crowd: Validating ‘Crowd’ Sources for Improved Situational Awareness in Disaster Response’, Procedia Engineering (Elsevier 2016)

Simpson E, Venanzi M, Reece S, and others, ‘Language Understanding in the Wild’ (Association for Computing Machinery (ACM) 2015)

Fischer J, Reeves S, Rodden T, and others, ‘Building a Birds Eye View’ (Association for Computing Machinery (ACM) 2015)

Jones D, Fischer J, Rodden T, and others, ‘Augmenting the Bird Table: Developing Technological Support for Disaster Response’, Procedia Engineering (Elsevier 2015)

Ramchurn S, Huynh T, Ikuno Y, and others, ‘HAC-ER: A Disaster Response System Based on Human-Agent Collectives (demonstration)’, Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS (2015)

Ramchurn S, Huynh T, Ikuno Y, and others, ‘HAC-ER: A Disaster Response System Based on Human-Agent Collectives’, Proceedings of the International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS (2015)

Ramchurn S, Wu F, Jennings N, and others, ‘AtomicOrchid: Human-Aqent Collectives to the Rescue’, 13th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2014 (2014)

Fischer J, Ramchurn S, Osborne M, and others, ‘Recommending Energy Tariffs and Load Shifting Based on Smart Household Usage Profiling’, IUI ’13 Proceedings of the 2013 international conference on Intelligent user interfaces (Association for Computing Machinery 2013)

Ramchurn S, Osborne M, Parson O, and others, ‘AgentSwitch: Towards Smart Energy Tariff Selection’, 12th International Conference on Autonomous Agents and Multiagent Systems 2013, AAMAS 2013 (2013)

Ramchum S, Osborne M, Parson O, and others, ‘AgentSwitch: Towards Smart Energy Teriff Selection’, 12th International Conference on Autonomous Agents and Multiagent Systems 2013, AAMAS 2013 (2013)

Simpson E, Reece S, Penta A, and others, ‘Using a Bayesian Model to Combine LDA Features With Crowdsourced Responses’, 21st Text REtrieval Conference, TREC 2012 (2012)

Reece S, ‘Self-Adaptive Multi-Sensor Systems’, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (2001)

Reece S, ‘Qualitative Model-Based Multisensor Data Fusion and Parameter Estimation Using Infinity-Norm Dempster-Shafer Evidential Reasoning’, Proceedings of SPIE - The International Society for Optical Engineering (1997)

Logan B, Reece S and Jones K, ‘Modelling Information Retrieval Agents With Belief Revision’, Proceedings of the 17th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 1994 (1994)

preprint (2)

Open

Lees T, Reece S, Kratzert F, and others, ‘Hydrological Concept Formation Inside Long Short-Term Memory (LSTM) Networks’

Duporge I, Isupova O, Reece S, and others, ‘Using Very High-Resolution Satellite Imagery and Deep Learning to Detect and Count African Elephants in Heterogeneous Landscapes’