Machine Learning methods for precision predictions in high energy physics
Lead Research Organisation:
Durham University
Department Name: Physics
Abstract
This project will investigate the use of cutting edge machine learning technology for use
in studying and modelling the experimental data currently being taken by the Large
Hadron Collider.
in studying and modelling the experimental data currently being taken by the Large
Hadron Collider.
Publications
Bullock Joseph
(2019)
Automated Speech Generation from UN General Assembly Statements: Mapping Risks in AI Generated Texts
in arXiv e-prints
Logar Tomaz
(2020)
PulseSatellite: A tool using human-AI feedback loops for satellite image analysis in humanitarian contexts
in arXiv e-prints
Badger Simon
(2020)
Using neural networks for efficient evaluation of high multiplicity scattering amplitudes
in arXiv e-prints
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
ST/P006744/1 | 01/10/2017 | 30/09/2024 | |||
1941883 | Studentship | ST/P006744/1 | 01/10/2017 | 30/09/2021 | Joseph Bullock |
Description | Informed United Nations High Level Panel on Hatespeech |
Geographic Reach | Multiple continents/international |
Policy Influence Type | Participation in a guidance/advisory committee |
Impact | The UN's high level panel on hate speech meets regularly to discuss the UN's stance on the growing global issue of hate speech. In building up to the public launch of the UN's formal agenda and stance on combatting hate speech, UN Global Pulse (where I was doing my secondment) was consulted on the draft of the high level panel's white paper. We noticed that social media and the dangers of automated text generation using modern artificial intelligence methods was missing from this and so produced an internal white paper in response, highlighting some of the dangers of this technology and the impact on society. This paper was accepted by the panel and subsequently the panel's agenda and stance now puts a great deal of weight on the growing issue of hate speech on social media and the potential negative impacts of technologies such as artificial intelligence. Our internal white paper that was shared with the panel has since been turned into a conference paper and presented at one of the largest machine learning conferences in the world. |
URL | https://arxiv.org/abs/1906.01946 |
Description | United Nations Global Pulse |
Organisation | United Nations (UN) |
Country | United States |
Sector | Public |
PI Contribution | I joined UNGP as a research fellow for a year, with 6 months based out in New York as part of my time allowed under the award. Through this fellowship I have worked closely with the UN on a variety of research projects. Throughout these research projects I have applied and developed techniques used in my physics research. Out of this fellowship I have currently co-authored 2 papers and have 2 more in the pipeline. |
Collaborator Contribution | UNGP offered domain expertise, office space, a collaborative environment and computing resources. The mentorship I have received from them has significantly developed me as a researcher and otherwise. UNGP also funded, or part-funded, several research trips to conferences around the world. |
Impact | 2 conference papers in world leading conferences, attendance at multiple conferences and events |
Start Year | 2018 |
Title | PulseSatellite |
Description | PulseSatellite is a satellite image analysis tool for humanitarian disaster relief that is designed to host and allow interaction with machine learning models specifically designed for certain analysis tasks. This web based tool promotes human-AI feedback loops for expert model improvement and validation. The work grew out of a secondment with United Nations Global Pulse. |
Type Of Technology | Webtool/Application |
Year Produced | 2020 |
Impact | The web based tool is currently being used by UNOSAT, the UN's satellite programme, to speed up their rapid response to flooding, as well as for the monitoring of refugee and IDP camps. At least 15 other UN agencies have requested access. |
URL | https://www.unglobalpulse.org/microsite/pulsesatellite/ |
Title | XNet |
Description | Code accompanying a publication designing a new methodology for the segmentation of X-Ray imagery into bone, soft tissue and open beam regions. |
Type Of Technology | Software |
Year Produced | 2019 |
Open Source License? | Yes |
Impact | The procedure of X-Ray image segmentation is a crucial step in many X-Ray image processing techniques. Previously, much of this segmentation was one by hand, or semi-automated techniques requiring a great deal of fine tuning. We developed a novel, automated technique that we hope requires little fine tuning in practice and publish the code to enable this. This tool has also already been integrated into the pipeline of proprietary software in an X-Ray imaging company. |
URL | https://arxiv.org/pdf/1812.00548.pdf |
Title | n3jet |
Description | A software release accompanying a publication on using neural networks for high precision scattering amplitude approximation. This software will allow other researchers to design, build, train and deploy neural networks for scattering amplitude approximations and reproduce our work. We hope the release of his source code will allow for the testing of our methodologies in practice to formally measure the practical implications of incorporating neural networks into event generators for speeding up the calculation of hard process matrix elements. |
Type Of Technology | Software |
Year Produced | 2020 |
Open Source License? | Yes |
Impact | By creating interpolation methods for high precision scattering amplitude approximation we hope to significantly increase the time required to calculate matrix elements for har scattering processes. This is one of the largest bottlenecks in LHC simulations. |
URL | https://arxiv.org/pdf/2002.07516 |
Description | National Science Festival |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Postgraduate students |
Results and Impact | Over 3000 people attend the national science festival held in Durham every year. I took part in helping demonstrate some of the research we undertake in the physics deparement both in 2017 and 2020. This entailed explaining and demonstrating our Galton Board and Dark Matter detector outreach activities. |
Year(s) Of Engagement Activity | 2017,2019 |
Description | Talk at school |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Postgraduate students |
Results and Impact | I gave the keynote talk at a large annual school event to students from ages 12-18. This talk was on my secondment work to the United Nations and demonstrated some of my research there. |
Year(s) Of Engagement Activity | 2019 |