Dr Georgios Leontidis

Biography

I am the University's Interdisciplinary Research Director for Data and AI, and a Scottish AI Alliance Leadership Member. I have a strong interest in both theoretical aspects of Machine/Deep Learning, e.g. capsule networks, domain adaptation, self-supervised learning etc., as well as applications, e.g. data imputation in environmental data of COSMOS-UK network (PI in NERC/EPSRC project ENTRAIN - NE/S016236/1, NE/S016244/1), homomorphic encryption with deep learning for enabling data sharing and analytics in food industry (PI - IoFT network plus EPSRC project), anomaly detection in nuclear reactors (Co-PI in H2020 project Cortex - 20 EU partners), Optimising retail refrigeration systems with machine learning (Co-PI-IUK project with Tesco), forecasting yield in strawberries and tomatoes (Co-PI-EU Interreg project SmartGreen and PhD studentship), Gas Turbine availability and fault prediction with Siemens Lincoln, etc. I am currently leading/co-leading three funded projects, i.e. Enhancing Agri-Food Transparent Sustainability (EP/V042270/1), Predictive Emissions Monitoring System for Gas Turbines with Siemens Energy (EP/W522089/1) and Machine Learning and Expert-based System for Soft Fruit Yield Forecasting (Data Lab and Angus Soft Fruits).

Previously I was a Senior Lecturer at the University of Lincoln, before that a Senior Data Scientist at IBA Dosimetry in Germany and a Marie Curie FP7-ITN Fellow.

I am serving as programme committe in various top venues, such as NeurIPS, ICML, AAAI, ICLR, etc. and participated in the UK AI Council’s Data Working Group ecosystem. I am also a member of the Full College of EPSRC and a panel college member of the UKRI FLF. I am also serving as External Examiner at Cranfield University (MSc applied AI). I have also served as external panel member for programme validation events.

I have successfully supervised PhD students to completion and have served as PhD examiner several times.

I am currently supervising 11 PhD students.

Prizes and Awards

- Shortlisted, AUSA/UoA for "Outstanding Contribution to Accessibility and Inclusivity in Blended Learning (2021)"

- Ranked at Top 4% of the EPSRC Full Peer Review College

- NeurIPS 2020, top 10% Reviewer out of ~7000

- EU commission FISA 2019 conference - best PhD paper award (PhD student:Aiden Durrant)

Research Overview

I am interested in problems revolving around deep learning and machine learning, more specifically on domain adaptation, variational inference and self-supervised learning. I am also working on novel neural network architectures, such as Capsule Networks.

In terms of application areas, I have a strong interest in problems that ML can provide solutions primarily in environmental, industrial, food and healthcare settings.

My past and current activity involves working with national and international collaborators on nuclear reactor perturbation analysis, optimising retail refrigeration systems, gap filling in environmental time-series, domain adaptation for food retail packaging image quality detection, disease detection, and yield forecasting for strawberries and tomatoes

Current Research

I am currently working on the following problems across a several funded projects:

a) Detecting various types of perturbations via neutron noise modelling and deep learning. We are using simulated and real data for various types of nuclear reactors. The data are provided by our EU partners (EU-H2020, 2017-2021, http://cortex-h2020.eu/)

b) New routing algorithms for Capsule Networks in order to improve their run time and performance, whilst reducing the number of parameters

c) Yield forecasting for strawberries - we use mobile robots to collect data in a setting that our collaborators at the Univeristy of Lincoln have in the Riseholme campus. We collect time-series, image, depth and video data, so that we can develop new machine learning techniques that can accurately and robustly forecast yield in 1-, 2- and 3- weeks ahead

d) Predicting availability of gas turbines, a collaboration with Siemens Energy Industrial Turbomachinery Ltd.

Past Research

--Gap filling in environmental time-series, specifically for the Cosmos-UK network. We developed new data imputation techniques in order to fill the gaps in the time series using historical data from most of the Cosmos-UK sites across the UK (https://www.ceh.ac.uk/our-science/projects/entrain) - project funded by NERC, EPSRC and Defra (NERC-led)

--Optimising demand side response of retail refrigeration systems with Machine Learning, a collaboration with Tesco and funded by Innovate UK

Collaborations

UK:

a) Centre for Ecology and Hydrology, Wallingford, with Matt Fry, Jon Evans, Steve Cole, Mike Bowes and John Wallbank

b) British Geological Survey, Keyworth, with Andy Kingdon and john Bloomfield

c) Sheffield University, Mike Mangan

d) University of Lincoln, MLearn group, LIAT and LCAS groups

e) Siemens Energy Industrial Turbomachinery Ltd.

International:

a) Chalmers University of Technology, Sweden with Christophe Demaziere and Paolo Vinai

b) Paul Scherrer Insitute, Switzerland with Hamid Dokhane

c) National Technical University of Athens, Greece with Andreas Stafylopatis and Georgios Alexandridis

d) Technical University of Madrid, Spain with Cristina Montalvo

e) Nuclear plant, UJV/Rez, Czech Republic, with Petr Stulik

Funding and Grants

• • Enhancing Agri-Food Transparent Sustainability (Uni of Aberdeen, with Universities of Nottingham and Dundee, and Scotland’s Rural College) – EPSRC – £1,1M – Co-I (work package leader) – 01/2022 to 12/2024
  • More real than reality: using deep learning and generative models to resolve how people make sense of other people’s behaviour – SGSSS-ESRC PhD – £70K – 10/2023 to 10/2027
  • Data for Net Zero (D4NZ) – Net Zero Technology Centre Ltd £1,06M – Co-I (work package leader on decision making) – 1/1/2022 to 01/05/2025
  • Machine Learning algorithms for microfluidic precision oncology assays – CENSIS – PI – £49,789 – 01/2023 to 01/2024
  • 20 Tenure-Track Interdisciplinary Fellows and 12 Interdisciplinary PhD studentships – Development Trust - £4M – Co-PI (jointly the five Interdisciplinary Directors, University of Aberdeen)
  • WYSA – NIHR consultancy on a protocol for AI application for mental health management – £50K (across 12 people) – 05/2022 to 05/2023
  • Machine Learning and Expert-based System for Soft Fruit Yield Forecasting – Data Lab Industrial Doctorate with Angus Soft Fruits – £66K – 07/2021 to 07/2024
  • Opening the black box: helping AI to persuade without bias – SGSSS-ESRC PhD – £70K – 10/2022 to 10/2026
  • Next Generation self-supervised Learning Systems for Vision Tasks – EPSRC HPC – PI – 44,640 GPU hours
  • iCASE EPSRC 4-year PhD studentship with Siemens Industrial Turbomachinery on Machine Learning for predictive emissions monitoring systems for gas turbine – PI - £118K (£89K EPSRC || £29K Siemens Energy)– 10/2021-10/2025
  • Data Trusts and Data Sharing in Food Supply Chains – EPSRC IoFT Network Plus – £12K – PI – (£50K in total) – 09/2020-03/2021
  • NEXTGEN: Neural-network Encryption; eXploration of Techniques for secure aGricultural data processing – EPSRC IoFT Network Plus - £28K (£50K in total) – PI – 03/2020 to 09/2020
  • Engineering Transformation for the Integration of Sensor Networks – NERC – £114K – PI – (£340K in total) – 02/2019 to 06/2020
  • Out of War Experiences: Hope for the Future (Metadata Analytics) – 165K£ (280K£ including match-funding) – EU H2020 – Co-I – University of Lincoln Coordinating – 2019 to 2022
  • BerryPredictor: Improving harvest forecasts, yield predictions and crop productivity by optimising zonal phytoclimates in covered strawberry production – Innovate UK – £80,000 – Co-I – 09/2019 to 08/2022
  • CORe monitoring Techniques and Experimental validation and demonstration – EU H2020 – £155K - Co-PI for Uni of Lincoln (~£5M in total) – 09/2017 to 08/2021 – http://cortex-h2020.eu/
  • SmartGreen–Big Data and Eco-Innovative resource use in the NSR Greenhouse Industry – EU Interreg – £530,000 (with 50% match funding) for UoL (~3M£ in total) – 09/2017 to 08/2021
  • The Development of Dynamic Energy Control Mechanisms for Food Retailing Refrigeration Systems – Innovate UK – 845,510£ - Co-I for Uni of Lincoln (~3.5M in total) – 09/2016 to 11/2018 - https://tinyurl.com/y9sj5tdp
  • ReACT Refrigeration AI Control Technologies – BBSRC seeding catalyst – £ 35K - Co-I – 10/2018 to 04/2019
  • Precision Agriculture: Machine Learning for Yield Prediction and Uncertainty Estimation - BBSRC-CTP-NPIF 4-year PhD studentship with NIAB - £99,300 - 10/2020 to 10/2024