Data Study Group - May 2023

Introduction

Introduction

Stage 1: Precursor Stage (part-time, online)

• The precursor stage will last one week in the run up to the 'event stage' (Monday 15 May – Friday 19 May 2023).
• The maximum time commitment is 2.5 hours a day.
• This includes online workshops, presentations and team building which will prepare participants for the 'event stage'.

Stage 2: Event Stage (full-time, in person)

• The 'event stage' will run at The Alan Turing Institute over one week (Monday 22 May – Friday 26 May 2023).
• Group work begins and continues throughout.

Applicants should be able to commit to the duration of the event. The Alan Turing Institute is committed to supporting individual circumstances, please do not hesitate to email [email protected] to discuss any reasonable adjustments.

Challenges

The challenges are:

NBN Trust

Spatiotemporal analysis of species in the UK using the largest publicly accessible source of biodiversity data

This challenge presents an exciting opportunity for researchers to work on the UK’s largest publicly accessible source of biodiversity data to help ensure that this remarkable national resource has the greatest possible value and impact for nature.

The datasets on the NBN Atlas consist of records of species observations within the UK at a given location and time. While the datasets are rich and valuable, they are incredibly heterogenous in space, time and taxonomic coverage. This challenge seeks to examine the comprehensiveness of the biodiversity inventory and what the impact of records from individual data providers is on the coverage and completeness of the inventory. We ask, where are the gaps in data and how can we measure our progress towards a goal of a comprehensive biodiversity inventory?

Peak District National Park Authority

Using multi-spectral remotely sensed imagery to improve land cover classification

National Parks in the UK have been shaped by the longstanding interaction of natural and cultural forces, with no element of the landscape untouched by past or present human activity. This has created a rich diversity of habitats such as broad open moorlands, grasslands, enclosed farmlands and wooded valleys that exhibit variation in both their spectral and spatial properties. However there is currently no contemporary overview of land cover of National Parks in the UK.

This DSG’s objective is to develop techniques that are able to leverage the information of multi-spectral satellite images to improve automated land cover classification. The PDNP recently acquired detailed aerial and satellite imagery of the National Park, and has started to classify land cover based on visible imagery. However, other spectral bands outside the visible spectrum (such as infrared etc.) have the potential to drastically improve classification performance of particular classes.

If successful, this analysis would help quantify the effects of human activities and climate change on the National Park’s landscape, biodiversity and land use from the 1980s until today. Furthermore, an automated land cover classification protocol would enable National Parks in the UK to continue to monitor landscape changes in the future, and help target our resources to protect and enhance the landscape.

Environmental Investigation Agency

Identifying tiger stripe patterns

The Environmental Investigation Agency (EIA) has an existing image database of 158 unique individual tiger skins following encounters with illegal wildlife traders in physical markets and online. Each skin is linked to one or more traders. They also collect images of seized tiger skins and carcasses. New skins are periodically added to the database and manually cross-referenced to determine if they are duplicate images, or duplicate skins but in different images; either of which can reveal dynamics of interest to law enforcement.

This project seeks to develop a user-friendly tool for identifying individual tigers from their stripe patterns, to inform enforcement efforts and counter trade in tiger skins, carcasses, and individuals.

Dstl

Topic Modelling to find topics and trends in academic papers.

The Discovery Project at Dstl uses topic modelling to find emerging topics and trends in academic papers for futures analysts. We hypothesise that studying a time series of topic model runs would allow important dynamics to be uncovered that would otherwise remain unseen in a single run. By tracking how topic models change over time, we expect emerging technologies can be identified at a faster pace than manual scanning, therefore giving the UK a competitive edge in assessing these technologies' applicability to Defence and Security.

To help identify emerging technologies, the Discovery project aims to develop an algorithm that can flag significant differences between a time series of multiple topic model runs. Can machine learning be applied to locate these differences and find emerging, merging/splitting and converging/diverging topics? Further, can the method be extended such that future trends can be predicted?

Johnson Matthey

Meeting the Challenges of Sustainable Chemical Plant Operations: A Machine Learning Approach for Optimizing Renewable Energy Use and Transient Dynamics.

Chemical processes are inherently complex, with multiple process states and paths possible for a given setting. As the industry moves towards for sustainable operations, the use of renewable energy and feedstocks is a key priority. As a consequence of this, chemical plant operations will depart from conventional steady-state operation (slow dynamics) to a more transient basis (fast dynamics) to maximise the use of the resource (e.g. wind power). Such a change introduces process control challenges and the use of large amounts of time-series data to ensure safe and reliable operation.

This challenge is to develop a machine learning-based modelling framework that can accurately predict and optimise plant response in subsequent time steps based on these time-series data.

Our early work in this area, as shown through a feasibility study on a Methanol simulator, has demonstrated that flexible Bayesian regression techniques can effectively optimise nonlinear processes at steady-state.