The Sigma Data Challenge was formally launched at our Technical Workshop, London, 11th October 2022
The Sigma Data Challenge is a pilot project to trial access for academic researchers to large scale data from multi-site radiation detectors. It’s an opportunity for researchers to develop machine learning or AI algorithms to analyse this large data set of spectra, and to make conclusions about the distribution of radiation sources that were present in the field.
- The Sigma Data contains up to 1.5 billion gamma spectra, spread across 3 data pilot data sets. The spectra are typically very sparse and contain few counts.
- Academic researchers will be given access to the data sets and invited to develop automated analysis methods that can interpret this data.
- Some geographical contexts will be given to the data, however the full ‘ground truth’ of the location and type of radioactive sources present in the field is not available.
- A key outcome of the project will be for the academic teams to give feedback to AWE on the quality of the data, the performance of their analysis algorithms, and recommendations for how the data could be improved in future trials.
The vision behind this Sigma Data Challenge is to provide access to these data, engaging the academic community in the development of distributed RN detection spectroscopic ID algorithms. The datasets which will be made available were obtained during the Sigma London pilots of 2017/2018 where almost one billion gamma ray spectra were recorded by around one hundred distributed detectors located in central London. The data were collected on three occasions, each spanning weeks to months in duration. The middle pilot data will be released first, and the intent is to respond to feedback from the community in subsequent calls by extending the datasets, for example by generating synthetic threat spectra to go with the as-collected spectra, as well as by releasing further datasets in following challenges.
Participation and access to the data will be by registration through NuSec, including agreement with terms and conditions and the expectation to participate in a NuSec Workshop in Autumn 2023.
UPDATE Jan 2023* - Six Academic/Industry groups have now taken up the Challenge and participated in an online workshop at the end November 2022 to discuss next steps for releasing and analysing the SIGMA data. Participants were also encouraged to collaborate and share information with each other during their planned research.
NuSec has invited Academic participants to apply for funding to pay for an undergraduate or postgraduate researcher (2-3 months) to support their proposed research project.
Participants are expected to report their findings at a NuSec Technical Workshop in the Sept/Oct 2023. It is hoped that the Challenge will lead to the production of papers and highlight issues for other future data analytic challenges.
If you are interested in participating in the Challenge, please read the presentation given at the NuSec Network Technical Workshop 11th October 2022 and complete the Application Form (includes terms and conditions) and submit to info@nusec.uk